Plasma processing apparatus and plasma processing method

Abstract

A plasma processing apparatus and method perform processing on a wafer as a processing target arranged in a processing chamber inside a vacuum container using a plasma formed in the processing chamber, in which light of a plurality of wavelengths is received from the surface of the wafer at a given plurality of times during processing of the wafer as the processing target, and in a case where the amount of processing during processing of the wafer as the processing target is detected using a result of comparing information indicating the intensity of the light received of the plurality of wavelengths with data indicating the intensity of the light of the plurality of wavelengths obtained beforehand, the similarity between each wafer is quantified based on data indicating the intensity of the light of the plurality of wavelengths from the surface of each wafer obtained beforehand in the processing of a plurality of wafers, and at least one data selected in correspondence with the quantified similarity is compared with data indicating the intensity of the light of the plurality of wavelengths obtained during processing of the wafer as the processing target to detect the amount of processing.

Description

TECHNICAL FIELD

[0001] The present application relates to a plasma processing apparatus or a plasma processing method. BACKGROUND

[0002] In the production of semiconductor devices, the following processes are performed: forming components that function as circuits that perform various functions as a whole on the surface of a semiconductor wafer, and wiring that connects a plurality of components to each other. By repeatedly performing processes including the formation of a film layer of various materials, such as a conductor, or a semiconductor or an insulator, that is formed in advance on the surface of a substrate-like sample such as a semiconductor wafer, and the removal of unnecessary portions of these film layers, and the like, the formation of these components and wiring is performed. In such a process of removing unnecessary portions, a process (a technique) using dry etching utilizing plasma is widely used.

[0003] In such etching utilizing plasma (also referred to as plasma etching), a gas for processing is introduced into a processing chamber provided inside a vacuum vessel of a processing apparatus, and a high-frequency electric field based on high-frequency power supplied from a high-frequency power source is supplied into the processing chamber, and atoms or molecules of the introduced gas are excited to be ionized or dissociated to be plasma, and the surface of a sample is exposed to and contacted with the plasma, and a reaction between particles in the plasma and a film layer of a processing target is induced and performed. At this time, anisotropic or isotropic etching of the film layer of the processing target is performed by physical reactions such as sputtering caused by charged particles such as ions in the plasma, chemical reactions caused by radicals (particles having a reaction activity, active species), and the like. By appropriately selecting and applying such processes each having different characteristics on the surface of a wafer, components and wiring having a configuration that functions as the above-described circuits are formed.

[0004] In a case where the processing shape of plasma etching is different from the design, the formed various components can not be able to achieve their functions. Therefore, many process monitoring techniques for monitoring / stabilizing the etching process have been proposed. For example, a process monitoring that measures the film thickness of a film formed on a wafer, the depth of a groove or a hole formed on a wafer, by measuring reflected light from the wafer in processing, is called film thickness / depth monitoring, and can be used in the end point determination of the etching process and the like.

[0005] In Patent Literature 1, a processing precision high-precision method utilizing this film thickness / depth monitoring is described. In this method, the etching process is ended at the timing when the film of the processing target is about to be completely removed, by using film thickness / depth monitoring using plasma light as a light source to detect this timing. Thereafter, the processing target portion and the processing non-target portion are etched by selectively switching the conditions for etching, the entire processing time is suppressed to be short, and there is no processing deviation in the wafer plane, and complete removal of the processing target film is achieved.

[0006] Further, in Patent Literature 2, a high-precision technique for the measurement precision of the film thickness / depth of the film thickness / depth monitoring is described. In this method, as a light source for irradiating a wafer, an external light source is used instead of plasma light. Thereby, the light quantity fluctuation of the light source becomes small, and high-precision measurement of the film thickness / depth is achieved. These film thickness / depth monitoring techniques previously acquire a pattern of reflected light from a wafer obtained in etching as a database (referred to as DB), and by comparing the reflected light of the wafer measured in etching with the DB, the processing state of the wafer is estimated. Therefore, in the case where the device structure of the wafer at the time of acquisition of the DB and the device structure of the wafer as an evaluation target are different, inconsistency occurs in the reflected light pattern of the DB and the reflected light pattern of the evaluation target, and there is a problem that correct film thickness / depth measurement cannot be achieved.

[0007] In view of this problem, in Patent Literature 3, a film thickness / depth measurement method corresponding to the device structure deviation between wafers as described above is disclosed. In this prior art, in the case where the film thickness of the base film of the etching target film is different for each wafer, the wafer reflected light patterns of the thick case and the thin case of the base film thickness are previously acquired as DBs, and by using these two DBs, correct film thickness / depth measurement is achieved for wafers of various base film thicknesses.

[0008] Prior Art Documents

[0009] Patent Literature

[0010] Patent Literature 1: JP Patent Publication No. 11-260799

[0011] Patent Literature 2: JP Patent Publication No. 2004-507070

[0012] Patent Literature 3: JP Patent Publication No. 2014-195005 SUMMARY

[0013] PROBLEMS TO BE SOLVED BY THE INVENTION

[0014] However, in the prior art of Patent Literature 3, the following problem occurs.

[0015] That is, in Patent Literature 3, in the case where the structure of the film on the wafer upper surface is known, data of a pattern using the wavelength of the intensity of interference light caused by the reflected light of a plurality of wafers in which the thickness of the base film of the etching target film is different for each wafer as a parameter is used as a database (DB), but in the case where the film structure of the wafer is unknown, correct film thickness / depth measurement cannot be achieved.

[0016] For example, in a case where there is a deviation in the thickness, shape of a film other than an etching target film of a film structure among a plurality of wafers, a plurality of test wafers having a film structure equivalent to such a wafer are processed in advance, data of interference light of reflected light is obtained as reference data, and from these data, a plurality of data are appropriately selected with respect to a processing depth, a remaining film thickness in processing of a wafer used in manufacturing of a semiconductor device, as a DB used for end point determination.

[0017] On the other hand, in a case where a film structure of a wafer is unknown, it is difficult to select appropriate data based on a difference in the structure among wafers. For example, in a case where a plurality of data arbitrarily selected are used to detect a remaining film thickness, a processing depth in processing, in a wafer having a structure different from a corresponding film structure, the precision is significantly reduced.

[0018] Therefore, there is a problem that if data of intensity of interference light of reflected light corresponding to all of a deviation in shape, size, material, and the like of a film structure of a generated wafer are not obtained in advance, a DB having such data cannot be used to detect a remaining film thickness, a processing depth with high precision. As such a deviation in film structure, not only a thickness of a base film of an etching target film, but also a thickness, size, and the like of a mask layer constituting a film structure, a shape, size, width, pitch of a groove, a trench of a film of a lower layer of an etching target film, a shape, size of a film layer around an etching target film, and the like, all factors that bring variation in a relationship between a remaining film thickness, a depth of an etching target film and intensity of interference light of reflected light from a wafer are included.

[0019] Further, there is a case where a film structure of a film causing reflected light is different depending on a detection position, range of reflected light on a wafer, and in this case, as well as the above, in processing of a wafer having a film structure different from a film structure corresponding to data of interference light of reflected light contained in a DB, there is a problem that detection of a remaining film thickness, a processing depth with high precision cannot be achieved using reflected light from the wafer. Further, in a case where a selection ratio of an etching target film and a surrounding material in etching processing is deviated among a plurality of wafers, a deviation also occurs in intensity of interference light caused by reflected light. In a case where data of reflected light corresponding to a plurality of film structures having such a deviation in selection ratio are not used, as well as the above, precision of detection of a remaining film thickness, a processing depth in processing is impaired.

[0020] Further, in a case where light is irradiated on a wafer from a light source arranged outside a processing vessel in order to obtain reflected light from the wafer, the same problem as described above occurs in a case where there is a deviation in the spectrum of the light source, a change in the intensity of plasma light formed in the processing vessel over time during processing. The cause of the problem is not only the film on the wafer surface but also the film that causes variation in the light, and it is difficult to obtain the pattern of such variation in advance. Therefore, even if the data of reflected light is obtained by actually processing a wafer in advance, since the variation in the light is less relevant between a plurality of wafers individually, the data involved in the reflected light cannot be appropriately selected, and the accuracy of detecting the remaining film thickness, the processing depth corresponding to the variation is impaired.

[0021] As described above, in the related art, in a case where information on the film structure on a wafer, the position, the range of reflected light, the deviation in the selection ratio of the etching object and other materials, the spectrum of the light source, the variation in the plasma light over time, and the like cannot be sufficiently obtained, the data of reflected light cannot be appropriately selected from the wafer surface corresponding to the deviations / variation, and a problem occurs in which the accuracy of detecting the amount of processing such as the remaining film thickness, the processing depth using the reflected light during processing of the wafer is impaired.

[0022] An object of the present application is to provide an plasma processing apparatus or a plasma processing method capable of detecting the amount of processing such as the remaining film thickness of a film layer of a processing object during processing of a wafer with high accuracy.

[0023] Means for solving the problem

[0024] To solve the above problem, one of the representative plasma processing apparatuses of the present application is achieved by the following, a plasma processing apparatus that performs processing on a wafer that is a processing object arranged in a processing chamber inside a vacuum container using plasma formed in the processing chamber, the plasma processing apparatus including: a light receiver that receives light of a plurality of wavelengths from the wafer surface at a given plurality of times during processing of the wafer that is the processing object; and a detector that detects the amount of processing during processing of the wafer that is the processing object using a result of comparing data indicating the intensity of the light of the plurality of wavelengths received and comparison data indicating the intensity of the light of the plurality of wavelengths obtained in advance, the detector numerically values the degree of similarity between each wafer based on data indicating the intensity of the light of the plurality of wavelengths from the surface of each wafer obtained in advance in the processing of a plurality of wafers individually, selects at least one data based on the numerically valued degree of similarity, and uses it as comparison data, compares the comparison data and data indicating the intensity of the light of the plurality of wavelengths obtained during processing of the wafer that is the processing object, and detects the amount of processing.

[0025] One of the representative plasma processing methods of the present application is achieved by processing a wafer as a processing target arranged in a processing chamber inside a vacuum container using a plasma formed in the processing chamber, in which plasma processing method, there are provided a measurement step of receiving light of a plurality of wavelengths from the surface of the wafer at a given plurality of times during processing of the wafer as the processing target; and a detection step of detecting an amount of processing during processing of the wafer as the processing target using a result of comparing data indicating intensities of the light of the plurality of wavelengths received with comparison data indicating intensities of the light of the plurality of wavelengths obtained in advance, in which detection step, degrees of similarity between respective wafers are quantified based on data indicating intensities of the light of the plurality of wavelengths from the surface of each of the wafers obtained in advance in the processing of the plurality of wafers, at least one data is selected based on the quantified degrees of similarity, and used as the comparison data, and the comparison data and data indicating intensities of the light of the plurality of wavelengths obtained in the processing of the wafer as the processing target are compared to detect the amount of processing.

[0026] Effects of the Invention

[0027] According to the present application, it is possible to provide a plasma processing apparatus or a plasma processing method capable of detecting an amount of processing of a film layer of a wafer as a processing target during processing of the wafer with high precision.

[0028] The above-mentioned problems, configurations, and effects will be clarified by the following description of embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 is a diagram schematically showing the configuration of a plasma processing apparatus according to an embodiment of the present application, (a) is a diagram showing the entire apparatus, and (b) shows the structure of a film thickness / depth calculation unit.

[0030] Figure 2 is a diagram schematically showing the structure of a film configuration of a wafer as a processing target according to the embodiment shown in FIG. 1, in which a plurality of layers including a film layer as a processing target are stacked on the upper surface of a semiconductor wafer, (a) shows the state before and after processing, and (b) shows the state in which the film thickness is deviated. Figure 1

[0031] Figure 3 is a graph showing an example of variation in the value of reflected light and the residual film thickness detected therefrom in the case where the wafer as the processing target having the film configuration shown in FIG. 2 is subjected to etching processing, (a) shows the relationship between the wavelength and the reflection intensity, and (b) shows the result of detecting the film thickness / depth and performing end point determination in the processing of a plurality of wafers. Figure 2

[0032] Figure 4 ​​is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 1 is a graph showing an example of the sum of the errors of the spectra of the reflected light detected in the processing of a plurality of wafers processed by the plasma processing apparatus according to the present embodiment shown in FIG. 1, (a) shows a relationship between the light intensity and the wavelength, (b) shows a relationship between the error and the wavelength, and (c) shows a relationship between the sum of the errors and the mask film thickness. Figure 2 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum.

[0033] Figure 5 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 1 is a graph showing an example of the sum of the errors of the spectra of the reflected light detected in the processing of a plurality of wafers processed by the plasma processing apparatus according to the present embodiment shown in FIG. 1, (a) shows a relationship between the light intensity and the wavelength, (b) shows a relationship between the error and the wavelength, and (c) shows a relationship between the sum of the errors and the mask film thickness.

[0034] Figure 6 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 1 is a graph showing an example of the sum of the errors of the spectra of the reflected light detected in the processing of a plurality of wafers processed by the plasma processing apparatus according to the present embodiment shown in FIG. 1, (a) shows a relationship between the light intensity and the wavelength, (b) shows a relationship between the error and the wavelength, and (c) shows a relationship between the sum of the errors and the mask film thickness.

[0035] Figure 7 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 1 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 2 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum.

[0036] Figure 8 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 7 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 2 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum.

[0037] Figure 9 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 1 is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 2

[0038] is a graph showing a pattern of the values of the remaining film thickness in the processing of the etching processing of the wafer having the film structure shown in FIG. 1, and the intensity of the reflected light from the wafer with a plurality of wavelengths as parameters, (a) shows a map, and (b) shows a film thickness spectrum. Figure 10 Figure 1 Figure 2 ​​A graph showing an example of a change in the value of the first derivative of the intensity of light of a specific wavelength under the intensity of light of a reflected light from a film structure of a wafer subjected to an etching process performed on a wafer of the film structure shown in FIG. 1.

[0039] Figure 11 is a graph showing an example of a relationship between the initial thickness of the mask layer of each wafer and the first and second principal component values. Figure 1 Another other modification of the plasma processing apparatus according to the embodiment shown in FIG. 1 is provided with Figure 2 A graph showing an example of the intensity of light of a reflected light of a plurality of wavelengths in a case where the film subjected to a process in an etching process performed on a plurality of wafers of the film structure shown in FIG. 1 is given a residual film thickness.

[0040] Figure 12 is a graph showing an example of a relationship between the initial thickness of the mask layer of each wafer and the first and second principal component values. Figure 1 Another other modification of the plasma processing apparatus according to the embodiment shown in FIG. 1 is provided with Figure 2 A plurality of data of the intensity of light of a reflected light from a film structure of a wafer subjected to an etching process performed on a plurality of wafers of the film structure shown in FIG. 1.

[0041] Figure 13 is a graph showing an example of a relationship between the initial thickness of the mask layer of each wafer and the first and second principal component values. Figure 1 Another other modification of the plasma processing apparatus according to the embodiment shown in FIG. 1 is provided with Figure 2 A graph showing an example of a relationship between the initial thickness of the mask layer of each wafer and the maximum value of the variance of the distance between each data obtained by using Isometric Mapping on a plurality of data of the intensity of light of a reflected light from a film structure of a wafer subjected to an etching process performed on a plurality of wafers of the film structure shown in FIG. 1.

[0042] Figure 14 is a table showing an example of a data table in which a value of an error of a given residual film thickness of another wafer detected using each of a plurality of data of the intensity of light of a reflected light from a film structure of a wafer is a component. Figure 1 Another other modification of the plasma processing apparatus according to the embodiment shown in FIG. 1 is provided with Figure 2 A plurality of wafers of the film structure shown in FIG. 1 are subjected to an etching process, and

[0043] Figure 15 is a graph showing an example of a relationship between the initial thickness of the mask layer of each wafer and the first and second principal component values. Figure 14A table of combinations of database wafers of which the remaining film thickness of n wafers selected from the data table is shown.

[0044] Figure 16 is a graph schematically showing the error of the given remaining film thickness of other wafers detected using the respective pieces of data of the intensity of light from the film structure of the wafer surface Figure 1 Yet another other modification of the embodiment shown is directed to a plasma processing apparatus that performs etching processing on wafers each having a film structure shown in Figure 2 is a top view of the position of the wafer surface from which reflected light from the wafer is reflected when a plurality of wafers each having a film structure shown in

[0045] Figure 17 is a graph showing the sum of the errors of the given remaining film thickness of other wafers detected using the respective pieces of data of the intensity of light from the film structure of the wafer surface Figure 1 Yet another other modification of the embodiment shown is directed to a plurality of plasma processing apparatuses that perform etching processing on wafers each having a film structure shown in Figure 2 is a graph showing the sum of the errors of the given remaining film thickness of other wafers detected using the respective pieces of data of the intensity of light from the film structure of the wafer surface

[0046] Figure 18 is a graph showing the sum of the errors of the given remaining film thickness of other wafers detected using the respective pieces of data of the intensity of light from the film structure of the wafer surface Figure 1 Yet another other modification of the embodiment shown is directed to a plasma processing apparatus that performs etching processing on wafers each having a film structure shown in Figure 2 is a graph showing the sum of the errors of the given remaining film thickness of other wafers detected using the respective pieces of data of the intensity of light from the film structure of the wafer surface

[0047] Figure 19 is a graph schematically showing the error of the given remaining film thickness of other wafers detected using the respective pieces of data of the intensity of light from the film structure of the wafer surface Figure 1 is a block diagram showing the outline of the structure of a system of a plasma processing apparatus to which the embodiment shown is directed. DETAILED DESCRIPTION

[0048] (SUMMARY OF THE EMBODIMENTS)

[0049] In the embodiments of the plasma processing method or the plasma processing apparatus of the present application described below, data showing the intensity of reflected light from the wafer surface or changes thereof when processing such as etching is performed on a plurality of wafers is obtained, the degree of similarity between the respective wafers is calculated as a numerical value on the basis of the data of the intensity of reflected light of each wafer. Further, using the calculated numerical values of the degrees of similarity, a plurality of data establishing the relationship between the intensity of light and the remaining film thickness or the depth of processing with the wavelength of light from the wafer as a parameter is selected or calculated, and using these data, detection of the remaining film thickness or the depth in processing is performed.

[0050] The plurality of wafers that acquire data of reflected light from the surface thereof is a plurality of (2 or more) wafers in a process of the same processing, for example, a wafer (product wafer) suitable for using a process of a semiconductor device that is implemented in mass production to become a product. Further, instead of using these product wafers, a test wafer that is processed under conditions equivalent to the conditions at the time of processing of the product wafer for the purpose of acquiring data of the reflected light described above can also be used. Alternatively, data based on a simulation that reproduces reflected light in the process of implementing the process, particularly, reproduces a deviation of the reflected light of the wafer generated in the process can also be used.

[0051] As data of the reflected light, a plurality of data (data set) that indicates a value of the intensity of the reflected light from the surface of the wafer in the process accompanying the passage of time as a whole is used. For example, the reflected light from the wafer in etching can be spectrally divided into each of a predetermined plurality of wavelengths at each time, and so-called time series data in which a signal indicating the intensity of the light detected for each wavelength is related as a signal at each time is used as data of the reflected light. Further, data indicating a change in the value of the intensity of the light for each wavelength of a predetermined wavelength or a given range obtained at each time can also be used as data of the reflected light.

[0052] As data used in the calculation of the degree of similarity between the wafers, at least the intensity of the light that reflects the remaining film thickness or depth of the film that is the object of the etching process can be indicated as an index. For example, as this data, a value of the amount of light detected from the surface of the wafer including the film layer of the object, data in which noise superimposed on this amount of light is removed or reduced by performing digital signal processing thereon, and corrected, can be used. In a case where a variation in the amount of light, a variation in the etching speed, or the like occurs in each of the plurality of wafers, this data can also be a result of normalizing the amount of light based on a specific amount of light.

[0053] The index indicating the intensity of the light can be data that is integrated into a form that enables comparison of the difference between each of the plurality of wafers. For example, in a case where the range of the value of the remaining film thickness in the etching process of each of the plurality of wafers, the speed of etching, is different, the amount of light detected for each wafer is converted into the remaining film thickness, and in a range in which the wafers become the same remaining film thickness, data of the amount of light can be used as an index indicating the intensity of the light. Further, in a case where the number of data of the film thickness detected in the processing of each wafer (the number of points of time in the processing in which detection is performed) is different, interpolation / resampling processing of the number of points can be performed using data obtained by interpolation such as spline interpolation, and data in which the number of data points is equalized can be used as an index indicating the intensity of the light.

[0054] In a case where the range or the number of wavelengths of the obtained reflected light differs among the wafers, the same processing as in the case of the different film thicknesses can be performed. In addition, as an index indicating the intensity of light, not only so-called two-dimensional data using the film thickness and the wavelength as parameters, but also one-dimensional data in which data at a specific residual film thickness or data at a specific wavelength is extracted can be used.

[0055] The calculation of the similarity among the wafers using the intensity of light reflected from the surface in the processing of each wafer is performed by calculating the value of the difference in the intensity of light between the wafers. For example, the average value of the intensity of light indicated by the two-dimensional or one-dimensional data obtained for all the wafers as the object of the calculation of the similarity is calculated, and the error absolute value or the error square with respect to the average intensity of light is calculated for the intensity of light data of each wafer, and the value of the sum of these is used as an index of the similarity. Alternatively, the cosine similarity of the intensity of light of the reflected light obtained from each wafer with respect to the value of the average intensity of light can be used.

[0056] Further, the principal component value calculated by principal component analysis of the intensity of light data of the entire wafer can be used as an index of the similarity. Furthermore, a value obtained by applying a dimension reduction method to the intensity of light data of the entire wafer can be used. Specifically, Isomap, LLE, Laplacian Eigenmap, Hessian Eigenmap, spectral clustering, diffusion map, kernel PCA, or the like can be used.

[0057] The plurality of data selected or calculated based on the value indicating the calculated similarity is obtained from a plurality of (two or more) wafers having different values of the similarity, and the wafers including the minimum and the maximum of the values of the numerical values or the wafers near them are selected. For example, the wafers including the minimum and the maximum of the values of the numerical values and having the numerical values sandwiched by them can be extracted at substantially equal intervals on the numerical value axis.

[0058] Further, the wafer range in which the film thickness / depth measurement can be performed can be calculated from the DBs of the wafers by setting each wafer in the DB and performing the film thickness estimation for the other wafers, the wafer combination in which the film thickness / depth measurement of the entire wafer can be performed is determined, and the plurality of DBs is selected using the wafer combination in which the number of wafers is the smallest.

[0059] Here, whether the film thickness / depth measurement can be performed can be determined based on whether the error between the estimation result and the measured result of the film thickness / depth is within an arbitrary target error. Further, with respect to the selection of the plurality of DBs from the determined wafer combination, the wafer combination in which the number of wafers is the smallest and the cumulative number of wafers in which the film thickness / depth can be measured is the largest can be selected. Furthermore, in a case where the wafer data including a clear abnormality is included, the wafer can be excluded from the determination of whether the film thickness / depth measurement can be performed.

[0060] The film thickness / depth determination method using the plurality of DBs set by these methods can be performed by comparing the reflection light data obtained from the evaluation target wafer with the plurality of DBs. For example, the reflection light of the evaluation target wafer is compared with each DB, and the estimated value of the film thickness / depth in each DB is calculated, and the estimated value of the film thickness / depth of the DB having the smallest matching error with the reflection light of each DB is used as the value of the film thickness / depth at the current time. The matching error is not limited to the error at the current time, and can be set to the total including the past time.

[0061] According to the embodiment of the present application, the reflection light data at the time of etching a plurality of wafers can be obtained, the similarity between the wafers is numerically evaluated based on the light intensity of the reflection light data of each wafer, the plurality of DBs are selected using the result of the numerical evaluation of the similarity, and the measurement of the film thickness / depth is performed using the selected DB.

[0062] A plasma processing apparatus and a plasma processing method for detecting a remaining film thickness or a processing depth and processing a wafer as a processing target will be described below with reference to the drawings according to an embodiment of the present application. In the present embodiment, a semiconductor manufacturing apparatus for performing etching processing on a wafer arranged in a processing chamber inside a vacuum container and having a unit for detecting a remaining film thickness / processing depth during processing will be shown, and a procedure for detecting a remaining film thickness or a processing depth during etching processing performed in the semiconductor manufacturing apparatus will be described.

[0063] [Embodiment 1]

[0064] The present embodiment will be specifically described below using Figures 1 to 6 Figure 1 is a schematic view showing the configuration of a plasma processing apparatus according to the present embodiment. In particular Figure 1 (a) of FIG. 1 is a schematic view showing a longitudinal cross-sectional view of a plasma processing apparatus according to the present embodiment.

[0065] In the plasma processing apparatus shown in the drawing, an etching gas introduced into the inside of a vacuum processing chamber 10 from a gas introduction unit (not shown) is excited and decomposed by power generated using a high-frequency power source (not shown) or the like, microwaves, and becomes plasma 12, and a processing target 16 such as a semiconductor wafer arranged on a sample stage 14 is subjected to etching processing (plasma processing) by the plasma 12.

[0066] ​The introduction of the gas into the vacuum processing chamber 10, the generation and control of the plasma 12, the voltage application to the processing target by a high-frequency power source or the like not shown, and the like are performed by the control section 40, and synchronization / timing adjustment between the respective devices is performed so that the desired etching processing is implemented. In the case where the plasma 12 is pulsed, the control of the pulsing is also performed by the control section 40. At this time, with respect to the plasma 12, the voltage application by the high-frequency power source or the like that ionizes the etching gas, the microwave irradiation, or the like is modulated to switch their on / off, and the plasma is pulsed. Further, the pulsing of the plasma is also performed by time-modulating the introduction of the etching gas.

[0067] The plasma processing apparatus is provided with a mechanism that measures the film thickness / depth of the processing target 16. Light emitted from the light source 18 is introduced into the vacuum processing chamber 10 via the optical system 50 and the introduction lens 20, and is irradiated to the processing target 16 as irradiation light 22. The light source section 18 uses continuous spectrum light continuous from ultraviolet to infrared wavelengths, but in the case where the film thickness / depth measurement is performed using a specific wavelength, a light source of the specific wavelength can be used. The reflected light 24 from the processing target 16 is introduced into the detection section 28 via the detection lens 26 and the optical system 50.

[0068] The detection section 28 is configured by a spectrometer that spectrally splits the introduced light, and detects the light amount of each wavelength. In the case where the film thickness / depth measurement is performed using a specific wavelength, the detector is not limited to the spectrometer, and a photodetector or the like can be used. At this time, if the light introduced into the detection section 28 is only the desired specific wavelength, a photodetector can be directly used, and in the case where continuous spectrum light is introduced, a mechanism that selects only the specific wavelength using a monochromator or the like can be provided in front of the photodetector.

[0069] Here, in (a) of the above-described embodiment, Figure 1 In the case of this structure, in order to most efficiently detect the reflected light 24, it is desirable to tilt the introduction lens 20 and the detection lens 26 so as to be directed to the same optical line with the processing target 16 as a reflecting surface.

[0070] The structure of the introduction lens 20 and the detection lens 26 is not limited to Figure 1 In the case of this structure, it is desirable to configure the optical line direction of the lens to be perpendicular to the processing target 16, and to be able to detect the perpendicular reflected light obtained as a result of the perpendicular irradiation. Further, in (a) of the above-described embodiment, Figure 1 In (a) of the above-described embodiment, one pair of the external light introduction system and the reflected light 24 detection system is described, but in the case where the film thickness / depth is measured at a plurality of positions of the processing target 16, a plurality of measurement systems can be provided.

[0071] In Figure 1 In (a) of FIG. 8, the case where light from the light source section 18 outside as a light source is incident is explained, but in the case where the plasma 12 is used as a light source, the light source section 18 can not be used. In the case where the plasma 12 is used as a light source, light emitted from the plasma 12 is also reflected by the processing object 16, and the reflected light 24 is detected as in the case of using the light source section 18. The data of the detection section 28 is introduced into the film thickness / depth calculating section 30, and the film thickness / depth is decided.

[0072] The database selection section 60 provides the database used in the decision of the film thickness / depth to the film thickness / depth calculating section 30. The database selection section 60, in the case where there are data of wafers that become candidates for a plurality of databases, numerically values the similarity of each candidate by comparing the light intensities of these wafer data. For example, the numerical value of the similarity is calculated using the absolute value of the error from the average value of the light intensity, the sum of the squares of the error, the cosine similarity, principal component analysis, Isomap, LLE, Laplacian Eigenmap, Hessian Eigenmap, spectral clustering, diffusion map, kernel PCA, and the like. The selection of the database is performed using the similarity of each wafer that is numerically valued.

[0073] The structure of the film thickness / depth calculating section 30 is explained using Figure 1 (b). FIG. 10 is a block diagram that divides the structure of the film thickness / depth calculating section 30 shown in (a) of FIG. 8 into blocks for each part that functions, and indicates the exchange, flow of data, information between each other with lines or arrows. Figure 1

[0074] As shown in the figure, the time series data Dl of the light amount of each wavelength introduced into the film thickness / depth calculating section 30 from the detection section 28 is removed, corrected for various noises, variations, and the like by the digital signal processing section 100, and is supplied to the waveform comparator 102 as time series data D2. The signal processing in the digital signal processing section 100 uses a low-pass filter in the removal of noise on the time axis of each wavelength. The low-pass filter can use, for example, a 2nd order Butterworth type low-pass filter, and the time series data D2 is calculated by the following formula.

[0075] D2(i) = b1 · D1(i) + b2 · D1(i-1) + b3 · D1(i-2) - [a2 · D2(i-1) + a3 · D2(i-2)]

[0076] ​Here, Dk(i) represents data of each data Dk at an arbitrary sampling time i, and the coefficients a, b are different in value depending on the sampling frequency and the cutoff frequency. Further, the coefficient values of the digital filter are, for example, a2 = -1.143, a3 = 0.4218, b1 = 0.067455, b2 = -0.013491, b3 = 0.067455 (sampling frequency 10 Hz, cutoff frequency 1 Hz).

[0077] In the case where removal of noise in a specific wavelength range of data Dk representing the intensity of light obtained at each sampling time is performed, the data Dk can be filtered by passing through a low-pass filter, or the S-G method (Savitzky-Golay-Method) can be used. Further, in the case where the light amount of each wavelength is removed from the data Dk, and the change (time change) in the light amount with respect to time is detected, signal processing that calculates the difference in the light amount (intensity of light) between the data Dk(i) of a plurality of times i, and the rate of change (derivative value) of the light intensity at each time i can be used. For example, using the data Dk(i) obtained at each sampling time i and the data of a given number of sampling times before and after, and applying the S-G method, the data Dk(i) is polynomialized and smoothed, and the derivative value is calculated, thereby obtaining time series data D2(i) of the derivative value of each of the sampling times i.

[0078] Such processing of data is processing called polynomial fitting smoothing differentiation, and is given by the following formula.

[0079] [Formula 1]

[0080]

[0081] Here, regarding the weight coefficients wj, in the case of 1st order differentiation calculation, for example, w-2 = -2, w-1 = -1, w0 = 0, w1 = 1, w2 = 2 are used. Further, in the case of 2nd order differentiation calculation, for example, w-2 = 2, w-1 = -1, w0 = -2, w1 = -1, w2 = 2 are used.

[0082] Further, in the data Dk(i) at an arbitrary sampling time, in the case where the values of the light amount of the object detected in the data change in time at the same rate for all wavelengths, processing that normalizes the values of the light amount of each wavelength by the average value of the light amount of all wavelengths, the sum of the absolute values of the values can be used.

[0083] In the present embodiment, the time series data D2(i) output from the digital signal processing section 100 is received at the waveform comparator 102, in which at least one pattern data representing a correlation of the film thickness / depth and the light amount of each wavelength, which is obtained in advance, is compared using an arithmetic unit, as data stored in the waveform pattern database 122. Here, the so-called pattern refers to a spectral pattern.

[0084] In the waveform comparator 102, the pattern data and the data D2(i) of each sampling time i of the time series data D2 are compared, and the pattern data having the smallest difference from the pattern of the data D2(i) is detected as the closest pattern data, among the pattern data of the light amount (intensity of light) of each wavelength at each film thickness or depth of processing or time after the start of processing, in which the pattern data establishes a correspondence between the values of the film thickness or depth of processing or time after the start of processing and the values of the intensity of light of the plurality of wavelengths as parameters.

[0085] As the data of the pattern having the smallest difference, for example, data in which the standard deviation between the plurality of wavelengths is the smallest can be used. The corresponding film thickness or depth of processing of the data of the closest pattern is calculated as the residual film thickness or depth of processing at the sampling time i. The value of the residual film thickness or depth of processing at each sampling time i calculated in the waveform comparator 102 is transmitted to the film thickness / depth storage section 104 and stored in a storage device such as a hard disk, a semiconductor RAM, a ROM, or the like, which is connected to the film thickness / depth storage section 104 so as to be able to communicate data.

[0086] The light amount data of each wavelength in the waveform pattern database 122 is data processed by signal processing implemented in the digital signal processing section 100. Here, in the case where there are a plurality of databases of pattern data of the film thickness / depth and the light amount of each wavelength in the waveform pattern database 122, there is a case where the film thickness / depth D3 determined using each database is supplied to the film thickness / depth storage section 104.

[0087] The film thickness / depth storage section 104 supplies the time series data D4 of the film thickness / depth to the optimum film thickness / depth decider 106.

[0088] In the optimum film thickness / depth decider 106, the optimum film thickness / depth is determined using the data supplied from the optimum database decider 124 and output to the outside of the film thickness / depth calculation section 30. For example, the film thickness / depth determined by the database number supplied from the optimum database decider 124 is output from the optimum film thickness / depth decider 106.

[0089] In this case, in the optimum database decider 124, data supplied from the waveform comparator 102 and / or the film thickness / depth storage section 104 is used to decide the optimum database. For example, data of a pattern for which a difference with another pattern is the smallest is determined as "optimum" data, and is selected as pattern data used in detection of the remaining film thickness / processing depth, where the pattern for which the difference with another pattern is the smallest is a result of pattern matching of a pattern closest to each database supplied from the waveform comparator 102 and a current pattern. In selection of the pattern data, not only data of the light intensity obtained at the current time i in processing, but also a total of differences obtained as a result of pattern matching of data of the light intensity at past times can be used. In this case, pattern data of the light intensity with respect to wavelengths at a plurality of times including the past times as parameters for which the difference becomes the smallest is selected as the pattern data.

[0090] Further, for example, a database for which a correlation coefficient of time and film thickness / depth is the smallest can be decided as the optimum database using time series data of the film thickness / depth of each database supplied from the film thickness / depth storage section 104.

[0091] The waveform pattern database 122 can use data including a plurality of data supplied from the fitted database calculator 120 as a database. For example, in the fitted database calculator 120, time series data Dl and / or time series data D2 and data supplied from the waveform pattern database 122 are used to generate a database including pattern data for which a value of a difference from a pattern indicating data of the light intensity obtained at the current time i coincides or a result of pattern matching becomes within a given allowable range, and is transmitted and supplied to the waveform pattern database 122. Further, for example, a database generated in the fitted database calculator 120 based on a predetermined operation can be supplied to the waveform pattern database 122. As the predetermined operation for calculating the database, a method of performing linear interpolation on two databases, an operation process of interpolating more than two databases with a polynomial can be used.

[0092] In Figure 1 In the film thickness / depth calculator 30 of (b), not only in a case where a plurality of databases exist in the waveform pattern database 122, but also in a case where only one database is used, the remaining film thickness or the depth of processing is detected from the intensity of the reflected light from the processing target 16 detected at an arbitrary time i in processing, and data indicating the same is output. In a case where the database stored in the waveform pattern database 122 is one, the fitted database calculator 120 and the optimum database decider 124 are not used, and data indicating a value of the film thickness / depth detected or calculated in the film thickness / depth storage section 104 is output from the film thickness / depth calculator 30 as it is.

[0093] Figure 1 The plasma processing apparatus shown in (a) uses a signal indicating the film thickness / depth output from the film thickness / depth calculation unit 30 to perform endpoint determination. That is, in the endpoint determiner that receives the signal from the film thickness / depth calculation unit 30, the value of the remaining film thickness or processing depth indicated by the signal is compared with a predetermined target film thickness or processing depth value. If it is determined to be within a given allowable range, the process is determined to have reached the endpoint; otherwise, if it is outside the allowable range, it is determined not to have reached the endpoint. If it is determined that the target remaining film thickness or processing depth has been reached, the arrival is announced by an announcer such as a monitor, lamp, or signal (not shown), and the control unit 40, which receives the signal indicating arrival, sends a signal to the plasma processing apparatus to stop the etching process or change the processing conditions.

[0094] In the plasma processing apparatus, the etching process of the object film layer on the surface of the object to be processed 16 is stopped based on the received etching stop signal, after the film thickness or processing depth has been detected, or the next processing step on the object to be processed 16 is performed after changing the processing conditions. Through this operation, the plasma processing apparatus of this embodiment can use the results of film thickness / depth detection to determine the endpoint.

[0095] use Figure 2 To illustrate the structure of the membrane, the membrane is constructed by stacking multiple layers of membranes on the surface of the object to be processed, namely the object 16, which is subjected to etching processing using the plasma processing apparatus described in the above embodiments to detect the membrane thickness / depth. Figure 2 This indicates that it is pre-configured in Figure 1 The illustrated embodiment is a schematic longitudinal cross-sectional view of a structure consisting of a multilayer film structure on the upper surface of a semiconductor wafer, which is the object being processed, and which contains the film layer of the object being processed.

[0096] like Figure 2 As shown in the left figure of (a), in the pre-processing film structure of the surface of the object to be processed 16, a base film 2 and a film to be processed 3 are overlapped in the vertical direction on the upper surface of a Si substrate 1 as the lower and upper layers, respectively, to form a film. Above the upper surface of the film to be processed 3, a mask 4 is disposed in the uncovered areas of the film layer of the object to be processed, consisting of a resin for forming a predetermined circuit pattern or a material of the film to be processed with a high selectivity.

[0097] As a result of etching such a film structure, such as Figure 2 As shown in the right figure of (a), in the processed film structure, a portion of the film 3 to be processed is removed. On the other hand, in the actual film structure that becomes the object of etching processing, as... Figure 2As shown in (b), there are deviations in the thickness of the mask 4. That is, for multiple processing objects 16, i.e. wafers, the mask layer before processing (in the early stage of processing) may be thinner (smaller thickness) or thicker (larger thickness) than the average thickness.

[0098] use Figure 3 This is to illustrate the impact of such mask film thickness deviation on the determination of the remaining film thickness / depth or endpoint of the processing of object 16. Figure 3 It is a schematic representation of having Figure 2 The graph shows an example of the reflected light obtained when multiple wafers of the film structure are etched, and the variation in the value of the remaining film thickness detected from it.

[0099] Figure 3 (a) is a graph showing the amount of reflected light from multiple wafers when the remaining film thickness of the processed object film on the surface of the wafer of processed object 16 is the same, but the initial film thickness of the mask is different. In this figure, the vertical axis represents the value when the amount of reflected light is transformed into wafer reflectivity. Figure 3 As can be seen in (a), due to variations in the mask film thickness, even if the remaining film thickness of the object being processed is the same, the magnitude of the reflectivity differs. That is, it can be seen that the distribution and profile (spectrum) of the intensity of reflected light at multiple wavelengths representing the remaining film thickness differ according to the mask thickness.

[0100] For multiple wafers of processing objects 16 with deviations in the initial film thickness of the mask for the aforementioned film structure, data is pre-acquired on the correlation between the patterns of values ​​representing the amount of multi-wavelength reflected light from the wafer at each sampling moment during processing of a single wafer and the remaining film thickness or processing depth at each of these detected moments. Using only this single data point, film thickness / depth is detected during the processing of the multiple wafers of processing objects 16, an endpoint determination is performed, and the results are... Figure 3 As shown in (b), the wafer has a film structure with a mask having an initial average film thickness.

[0101] In this example, the remaining film thickness, which is the target for endpoint determination, is set to 130 nm when multiple wafers with film thickness deviations have been etched. In the figure, the vertical axis represents the remaining film thickness, and the values ​​of the remaining film thickness measured by electron microscopy after etching of each wafer through destructive inspection are shown as dots in the figure.

[0102] As shown in this drawing, in a case where the end point is determined using the remaining film thickness or the processing depth detected using one pattern data obtained from the processing of one wafer that is taken in advance, the thickness of the film that is the processing target after the processing deviates from the target film thickness (130 nm) in each wafer, and it is known that the accuracy of the processing that forms the film structure based on the etching processing is impaired.

[0103] Thus, in a case where there is a large variation in the initial film thickness of the mask in each processing target 16, if the film thickness / depth is detected using data obtained from the processing of a specific wafer as an example, the accuracy of the end point determination and the accuracy of the processing are impaired, and further, the yield of the processing is impaired, and it can be impossible to achieve the manufacture of a semiconductor device with a high integration.

[0104] Therefore, in the plasma processing apparatus according to the present embodiment, the film thickness / depth is detected using a plurality of pattern data based on the reflected light obtained in the processing of the wafers of each processing target 16, and the end point of the processing is determined based on the detected remaining film thickness or processing depth. In the present embodiment, a plurality of pattern data relating to the data of the intensity of the reflected light of a plurality of wavelengths obtained from the wafer surface at a plurality of times in the processing and the remaining film thickness (or the time after the start of the processing) when the etching processing is performed on the same kind of film structure in which the film layers with the same material are stacked in the up-and-down direction are calculated in advance as numerical values. Figure 3 As shown in the example of (b) of the same, with respect to the etching processing when the wafers of a plurality of processing targets 16 of the same kind of film structure in which the film layers with the same material are stacked in the up-and-down direction are processed, a plurality of pattern data relating to the data of the intensity of the reflected light of a plurality of wavelengths obtained from the wafer surface at a plurality of times in the processing and the remaining film thickness (or the time after the start of the processing) are calculated in advance as numerical values.

[0105] Figure 4 (a) of the same is a graph that represents the relationship between the value of the intensity of the reflected light from the wafer with the remaining film thickness and a plurality of wavelengths as parameters in the processing of the etching processing performed on the wafer of the processing target 16 and the pattern.

[0106] In this drawing, the data representing the intensity of the reflected light detected in the processing is not detected in correspondence with the remaining film thickness, but is obtained as the spectrum of the reflected light of a plurality of wavelengths at each sampling time in the processing after the start of the processing. In the processing of converting each sampling time into the remaining film thickness, the value of the film thickness at each time is allocated by linear interpolation from the value of the film thickness at the initial stage (time 0) and the value of the film thickness at the final stage (the final time). The data of the graph shown in (a) of the same is obtained in each processing of each wafer. Figure 4 The data of the graph shown in (a) of the same is obtained in each processing of each wafer.

[0107] Therefore, in this embodiment, a common film thickness range for all the wafers is set using the aforementioned reflected light data obtained in advance regarding the wafers of the plurality of processed objects 16. Thus, within this film thickness range, pattern data representing the intensity of light at a given plurality of wavelengths for each film thickness of 1 nm is generated. Where the value of the pre-obtained data can be used for any film thickness, and where the film thickness is not present in the pre-obtained data, a value calculated by interpolation based on data regarding the preceding and following film thicknesses is used. For example, spline interpolation is used for the interpolation process.

[0108] Thus, for multiple wafers whose initial film thickness deviates from that of the mask for the film structure, interpolation processing is used to generate pattern data of reflected light with the wavelength in process as a parameter. From these generated pattern data, data corresponding to the wafer with the smallest actual remaining film thickness after processing is extracted when using any one identical pattern data for endpoint determination. Using this data as a benchmark, the pattern data corresponding to other wafers is compared, and the similarity between wafers based on the pattern data is calculated.

[0109] exist Figure 4 (b) shows an example of data (the spectrum of the film thickness) on the intensity of light calculated in this way with the wavelength as a parameter at a certain remaining film thickness. Figure 4 The film thickness spectrum of (b) is Figure 4 The spectrum of a specific film thickness in (a).

[0110] Figure 5 It means based on Figure 1 This diagram illustrates an example of the sum of the differences between the spectra of the film thickness obtained from the reflected light detected during the processing of multiple wafers by the plasma processing apparatus according to this embodiment. Figure 5 (a) Overlapping view shows the spectra of a given film thickness obtained from multiple wafers, each. As shown in this figure, the spectra of the film thickness corresponding to each wafer exhibit different values ​​at multiple wavelengths. Figure 5 Figure (b) shows the difference (error) between the spectral values ​​of each film thickness at each wavelength and their average value. In this figure, the similarity between the wafers is shown as a numerical value, representing the magnitude of the error.

[0111] Furthermore, the sum of the absolute values ​​of the errors across the entire wavelength are calculated for each chip. Figure 5 Figure (c) shows the results of establishing the correspondence between these values ​​and the initial film thickness of the mask for each wafer. The vertical axis in the figure represents the value of the total error, and the horizontal axis represents the film thickness of the mask (initial remaining film thickness) before the start of the etching process for each wafer.

[0112] In this example, it is assumed that the magnitude of the deviation of the specifications of the film layers constituting the film structure of each wafer, such as the deviation of the initial film thickness of the mask, is unknown, but in Figure 5 The value of the similarity calculated for each wafer described above is shown to correspond to the initial film thickness of the mask in (c) of FIG. 10. In this figure, for convenience, the value of the initial film thickness of the mask of each wafer and the value of the sum of the errors of each of a plurality of wavelengths in the film thickness spectrum of that wafer are established to correspond to each other to be plotted, and the sum of the errors of a given plurality of wavelengths in the film thickness spectrum of each wafer indicates a high correlation with the initial film thickness of the mask of that wafer. From this figure, it is known that the similarity of the film thickness spectrum of each wafer can be used to detect the magnitude of the difference in the structure between the film structures of the wafers.

[0113] Next, the plurality of wafers are sorted into a given order and are sequenced with a symbol or a number attached. For example, the plurality of wafers are sequenced in order from small to large according to the value of the sum of the errors calculated for each wafer. In Figure 6 The relationship between the number of each wafer and the sum of the errors of the plurality of wafers sequenced by such sorting is shown in (a) of FIG. 11. As described above, since the sum of the errors has a high correlation with the initial film thickness of the mask, in this example, the order in which the smaller the number of the wafer, the thicker the initial film thickness of the mask, and the larger the number, the smaller the initial film thickness of the mask, is attached.

[0114] In this embodiment, the wafer with the largest number, the wafer with the smallest number, and the wafer having a number between these wafers and having a sum of the errors equal to the difference between the sum of the errors of the wafer with the largest number and the sum of the errors of the wafer with the smallest number are selected from the plurality of wafers sequenced in this way, and three data of the pattern of the intensity of the reflected light with the wavelength as the parameter calculated from the above-mentioned data as needed using interpolation corresponding to these three selected wafers are used as the database. By selecting the data of the pattern of the spectrum of the reflected light used as the database in this way, the data of the reflected light corresponding to the wafer with the smallest and the largest initial film thickness of the mask having a deviation can be used as the database.

[0115] The comparison data is selected using the database obtained in this way, and the film thickness / depth of the wafer is detected by comparing the comparison data and the measured data, and Figure 6 The result of the end point determination based on the result thereof is shown in (b) of FIG. 12. In this example, by Figure 1 The pattern data with the smallest matching residual to each database is selected by the optimum database decider 124 shown in (a) of FIG. 13. Further, in this example, the plurality of wafers having a deviation in the initial film thickness of the mask are used as in (b) of FIG. 13. Figure 3

[0116] ​As is apparent from the graph, the remaining film thickness after processing is within ±0.5 nm of the target value of 130 nm in all of the wafers, and the processing target film layer can be processed with high accuracy. From this result, it is clear that according to the embodiment described above, even when the size, shape, material, and the like of the film structure on the wafer including the processing target film layer are unknown, the processing amount related to the remaining film thickness, processing depth, and the like during processing (hereinafter, referred to as processing amount) can be detected with high accuracy, and the end point of processing can be determined with high accuracy using the same.

[0117] The variation in the shape, size, material, and the like of the film structure between the plurality of wafers is one of the sources of stress in detecting the processing amount during processing, and the embodiment is not limited to the variation described above. The embodiment can be applied to a case where the variation in the processing conditions such as the variation in the characteristics of the film structure such as the base film thickness of the etching target film, the groove width, the depth, the structure of the film below the processing target film, the structure of the periphery, the variation in the position and range of the reflected light, the variation in the selection ratio of the materials of the etching target film and other films, the variation in the light source spectrum, the variation in the temporal change of the plasma light, and the like is generated between the plurality of wafers. Furthermore, it is obvious that the index of the light intensity, the data shaping of each wafer, the signal processing, the numerical value method of the wafer similarity, and the selection method of the plurality of databases of the embodiment are not limited to the above.

[0118] In the embodiment, the data of three wafers selected from the plurality of wafers in series is used as the database, but the number of databases is not limited to three. For example, it is obvious that the same effect as the embodiment can be obtained by using three or more data of the pattern of the intensity of the reflected light of the wafer of the maximum and minimum of the serial number and the number of the wafer divided at substantially equal intervals from these numbers as the database. Furthermore, it is obvious that the same effect as the embodiment can be obtained by using the data of a plurality of wafers having the maximum value and the minimum value of the sum of the errors of the wafers in series and the values of the sum of the errors divided at substantially equal intervals from these values in the database.

[0119] The following describes an example where the value of the change (temporal change) in the intensity of the reflected light of a specific wavelength from the wafer with respect to the passage of time is used as the numerical value of the similarity of each wafer. In this example, the determination of the end point of processing of the processing amount of the wafer detected using a plurality of data during processing is also performed. The conditions other than these are assumed to be the same structure as the embodiment 1 described above.

[0120] Figure 7 is a graph showing the relationship between the sum of the errors of the wafers in series and the similarity of the wafer. Figure 1 A modified example of the plasma processing apparatus related to the embodiment shown in FIG. 8 is a plasma processing apparatus 1 shown in FIG. 9. Figure 2The graph shows the amount of light-related quantities from the wafer surface obtained when the film structure is processed. Figure 7 Figure (a) is a graph illustrating, for example, the temporal variation of the intensity of a specific wavelength of light reflected from the wafer surface during wafer processing. In this figure, the remaining film thickness during processing is set as the horizontal axis to represent the change in light intensity, serving as the parameter representing the time variation.

[0121] Similar to the above-described implementation, film thickness allocation and film thickness range determination are performed on the data representing the spectrum of reflected light from each wafer obtained during the processing of multiple wafers, and interpolation processing is performed on the spectral data. Figure 7 (a) shows an example of the variation in light intensity at a specific wavelength based on the obtained interpolation data. It can be seen that the light intensity varies oscillatoryly with respect to the film thickness.

[0122] Figure 7 (b) indicates relative to Figure 7 (a) is a graph illustrating an example of the relationship between the time-varying intensity of reflected light at a specific wavelength, the average error of each wafer across the entire wafer, and the mask film thickness. Similar to the inter-wafer spectral comparison at a specific film thickness in the above embodiment, the time-varying intensity of light at the same wavelength for each wafer is extracted. Figure 7 Figure (b) shows the result of the sum of squared errors calculated for each wafer from the average value. As can be seen from the figure, when the time variation of light intensity is utilized, the sum of errors in each wafer also shows a correlation with the height of the mask film thickness, which can clarify the differences in construction based on the similarity of the spectra of each wafer.

[0123] Multiple chips are serialized by assigning them numbers in ascending order of the sum of their errors. Figure 8 The results show the relationship between the wafer number and the sum of the errors. For example... Figure 7 As shown in (b), the sum of errors is highly correlated with the mask film thickness; the smaller the wafer number, the larger the mask film thickness, and vice versa. Therefore, in Figure 8 The diagram shows that the sum of the errors increases in the same order according to the chip number.

[0124] In this example, using the results of such serialization, the wafers with the largest and smallest serial numbers are selected from a plurality of wafers, along with the sum of the errors between them. Figure 8the three wafers become equal or approximately equal to the extent that the intervals can be considered equal, the data of these selected wafers are used as data of the waveform pattern database 122. By using such data, the detection of the processing amount using the reflected light from the wafer is performed with high accuracy, corresponding to the minimum value and the maximum value of the initial mask film thickness that deviates from the value.

[0125] As a result of the end point determination performed based on the result of the detection of the processing amount in the processing using the database configured using the plurality of data thus selected and determined, the same result as that of Embodiment 1 Figure 6 (b) of the same, the film thickness after the processing is near the target 130 nm in all the wafers, and the error is also 0.5 nm or less. Therefore, it is apparent that in the detection of the processing amount using the plurality of data in the present embodiment, the detection of the processing amount is performed correctly even when the configuration deviation is unknown, and a high-accuracy end point determination is achieved.

[0126] Next, an example of a case where the two-dimensional data (plot) of the remaining film thickness of the processing target included in the film configuration on the wafer and the plurality of wavelengths of the reflected light from the film configuration is used as a parameter for characterizing the similarity of each wafer that has been processed in advance as a numerical value is described. As for the conditions other than the calculation of these similarity parameters for the plurality of wafers, the same structure as that of the above-described embodiments, Figure 7 , 8 the same structure as that of the above-described embodiments.

[0127] As in the above-described embodiments, in Figure 9 An example of the interpolation processing data of the film thickness, the wavelength, and the intensity obtained by performing the allocation of the film thickness, the determination of the film thickness range, and the interpolation processing of the spectral data on the data of the spectrum of each of the plurality of wafers that have been processed is shown. Figure 9 is a graph showing the relationship between the value of the remaining film thickness in the processing of the etching processing performed on the wafer having the film configuration described in Figure 1 the plasma processing apparatus related to the above-described embodiments and the value of the intensity of the reflected light from the wafer with the plurality of wavelengths as a parameter as a plot. Figure 2 The data shown in the present figure is the same graph as the plot cut out from a part of the plot of (a) of

[0128] According to the data of the present figure, as in the comparison of the spectra between the wafers at the specific film thickness related to the above-described embodiments, as a result of the calculation of the sum of the absolute values of the errors of the wafer average values of the interpolation processing data and each wafer, the same correlation of the mask film thickness and the error amount as that of (c) of Figure 4 Figure 4 Figure 5

[0129] ​​​Therefore, by using such an error amount, the same waveform pattern database 122 as (a) of Figure 6 can be determined, and the same as (b) of Figure 6 , high-precision film thickness estimation can be achieved with respect to mask deviation. Therefore, in this example, it is obvious that, even in a case where the deviation of the characteristics of the film structure including the film to be processed is unknown, correct processing amount detection can be performed, and high-precision end point determination can be achieved.

[0130] Next, an example of end point determination based on the result of detecting the processing amount in processing using data indicating the intensity of light of reflected light obtained by performing signal processing such as low-pass filtering, differential value calculation, and light amount normalization on data of reflected light of each of a plurality of wafers that have been processed in advance, based on data on which the signal processing has been performed, will be described. In this example, conditions other than the above-described signal processing are assumed to be the same as in the embodiment or modification example shown in Figures 1 to 9 .

[0131] As in the embodiment, the corresponding establishment of the residual film thickness and the determination of the residual film thickness range are performed with respect to spectral data obtained at a plurality of sampling times in processing of each wafer, interpolation processing is performed on the spectral data, and the amount of light of a plurality of wavelengths is normalized with respect to the average value of the amount of light of a plurality of wavelengths at each time of the data thus obtained. Further, for each wafer, the change in the intensity of light of a specific wavelength with respect to the change in the residual film thickness is extracted from the data obtained by performing the normalization, and the first differential value is calculated in the time direction using an LPF and an S-G method with respect to the time change in the amount of light of each wafer that has been extracted. In Figure 10 , an example of a graph showing the result thereof is shown.

[0132] Figure 10 is a graph showing an example of the change in the first differential value of the intensity in light of a specific wavelength with respect to the change in the residual film thickness of the film to be processed in processing of etching a wafer having the film structure shown in Figure 1 . Figure 2 . Figure 10 is a graph showing the time change in the amount of light of a specific wavelength related to reflected light, which is the same as (a) of Figure 7 . As a result of calculating the sum of the error squares of the whole wafer average value of the differential value of the amount of light and each wafer of the shape obtained by differentiating with respect to the film thickness direction of (a) of Figure 7 , the same error sum and mask film thickness correlation as (b) of .

[0133] Therefore, by using this error amount, it is also possible to select the data of the same waveform pattern database 122 as Figure 8 the result of which is known to be close to Figure 6 (b) of FIG. 8. Therefore, in this example, it is apparent that the correct amount of processing can be measured even when the structure deviation is unknown, and high-precision end-point determination can be achieved.

[0134] Next, an example in which the amount of processing in processing is detected using data representing the intensity of light of reflected light obtained by characterizing the similarity between a plurality of wafers each of which has been pre-processed using data of reflected light from the wafer on which data shaping with respect to the wavelength axis has been performed as a value, and end-point determination is performed using the result thereof, will be described. In this example, the conditions other than the above-described signal processing are assumed to be the same as those of the embodiment or modification example shown in Figures 1 to 9

[0135] In this example, for data of the spectrum of reflected light obtained at each sampling time in processing of each of a plurality of wafers each of which has been processed in the same manner as in the above-described example, the remaining film thickness is made to correspond thereto, and the determination of the range of the remaining film thickness, the interpolation processing in the time axis (film thickness axis) direction of the spectrum data is performed. Further, with respect to the wavelength axis, the data in the wavelength direction is also subjected to interpolation processing so that the wavelength scale becomes 5 nm in the wavelength range of 240 to 840 nm, and the data is reduced. In Figure 11 an example of spectrum data corresponding to the thinnest remaining film thickness of an arbitrary one of a plurality of wafers is shown as a result thereof.

[0136] Figure 11 is a graph showing an example of the intensity of reflected light of a plurality of wavelengths in the case where the film of the processing target obtained in processing of etching processing performed on a wafer having the film structure shown in Figure 1 is another other modification example of the plasma processing apparatus related to the embodiment shown in Figure 2 is a graph showing an example of the intensity of reflected light of a plurality of wavelengths in the case where the film of the processing target obtained in processing of etching processing performed on a wafer having the film structure shown in Figure 4 is a shape in which a part of the spectrum in (b) of FIG. 8 is cut out.

[0137] In this example, the data of this spectrum is also used to sequence the similarity of the structure between each of a plurality of wafers and each other in the same manner as in the above-described embodiment, and data selection is performed, and the remaining film thickness of the wafer in processing is detected using the selected data, and end-point determination is performed. As a result thereof, the deviation of the remaining film thickness of the wafer after processing can be obtained in the same manner as in the above-described embodiment. Therefore, in this example, it is apparent that the correct amount of processing can be measured even when the structure deviation is unknown, and high-precision end-point determination can be achieved. ​

[0138] Next, an example will be given of using principal component analysis (PCA) values ​​as parameters to represent the similarity between pre-processed wafers numerically. Aside from demonstrating the use of PCA for wafer similarity, the following will be assumed to be consistent with the above. Figures 1 to 12 The implementation and variations shown have the same structure.

[0139] Similar to the above implementation, for the data representing the spectra of reflected light obtained at multiple sampling times during the processing of multiple wafers, the corresponding establishment of the residual film thickness of the processing object, the determination of the range of the residual film thickness, and the interpolation processing of the spectral data are performed. Spectral data of reflected light corresponding to the same residual film thickness are extracted from each wafer. Principal component analysis is performed using the extracted spectra of each wafer. Figure 12 (a) and (b) show the relationship between the initial film thickness of the mask layer of each wafer and the first principal component and the second principal component.

[0140] Figure 12 It means to Figure 1 Another variation of the plasma processing apparatus according to the illustrated embodiment is provided for the use of... Figure 2 A graph illustrating the relationship between the first and second principal component values ​​obtained by principal component analysis of the intensity data of multiple lights in the reflected light from the film structure on the wafer surface obtained during the etching process of multiple wafers of the shown film structure and the initial thickness of the mask layer of each wafer.

[0141] From these diagrams, it can be seen that Figure 12 The first principal component value shown in (a) Figure 12 The second principal component values ​​of (b) all show a high correlation with the initial film thickness of the mask layer. Furthermore, the first principal component values ​​vary as a quadratic function containing a minimum and increasing / decreasing value relative to the initial film thickness of the mask. Therefore, from the representation having Figure 12 The first principal component value of the spectral data of the reflected light from multiple related wafers, as in (a), cannot uniquely determine the initial film thickness of the mask layer.

[0142] on the other hand, Figure 12The change of the 2nd principal component value shown in (b) with respect to the initial film thickness of the mask changes as a 1st order function with a substantially fixed slope. Accordingly, the correspondence between the 2nd principal component value shown in the graph and the initial film thickness of the mask layer can be used to detect the initial film thickness of the mask layer of each of the plurality of wafers and the similarity of the wafers based thereon. Thus, in this example, the 2nd principal component value is calculated from the data of the reflected light obtained at each sampling time in the processing of each of the plurality of wafers, and the similarity between each of the wafers based on the data obtained thereby is used to sequence the plurality of wafers. Further, data representing the pattern of the intensity of the light of the plurality of wavelengths of the reflected light is selected based on the result of the sequencing, and the data of the selected wafer is used as the data of the waveform pattern database 122.

[0143] In this example, using the plurality of data of the waveform pattern database 122, the amount of processing at each sampling time in the processing is detected from the data representing the intensity of the light of the reflected light from the wafer obtained in the processing of an arbitrary wafer, and the determination of the end point of the processing is implemented. As a result, the wafer to be processed is selected as the wafer to be processed, and the data of the selected wafer is used as the data of the waveform pattern database 122. Figure 6 As with (b), the residual film thickness after the processing is near the target 130 nm in all of the wafers, and the error is also 0.5 nm or less. Thus, in this example, it is also apparent that, in the case where the deviation in the characteristics of the film structure is unknown, the correct detection of the amount of processing can be performed, and a high-precision end point determination can be achieved.

[0144] Next, an example in which the similarity of the data of the pre-processed interference light from the plurality of wafers is numerically evaluated using manifold learning in the case where there is a deviation in the initial film thickness of the mask layer of the film structure constituting the wafer and the film thickness of the substrate disposed below the film to be processed will be described. In this example, the conditions other than the point that the similarity is numerically evaluated using manifold learning are also assumed to be the same as in the embodiment and the modified example shown in Figures 1 to 12

[0145] In this example, as with the example described above, the data representing the spectrum of the reflected light obtained at a plurality of sampling times in the processing of each of the plurality of wafers is processed to establish the correspondence of the residual film thickness of the object to be processed, determine the range of the residual film thickness, and interpolate the spectrum data, and the data of the spectrum of the same residual film thickness is extracted from each of the wafers.

[0146] Further in this example, the similarity of each of the wafers is numerically evaluated using the non-linear dimensionality reduction method of manifold learning, i.e., isometric mapping, using the extracted data of the spectrum. In isometric mapping, the spectrum of each of the wafers is taken as each data point, and the proximity relationship of each data point based on the K-nearest neighbor algorithm is calculated as a value using the data points.

[0147] ​Next, the geodesic distance between each data point on the K-neighbor chart is calculated using the calculated values of the proximity relationship, and a geodesic distance matrix is created using the values of these geodesic distances as components. The created geodesic distance matrix is subjected to multi-dimensional scaling (MDS) to project each data point to a low-dimensional space. Through the above process, each wafer is mapped to a low-dimensional space based on the similarity of the data of the corresponding spectrum, and a plurality of data of the database used for estimating the amount of the processing for using the mapping result, such as the residual film thickness, can be selected.

[0148] In this example, the data of the spectrum of the reflected light of each of a plurality of wafers that were pre-processed using isometric mapping is mapped based on the similarity. Furthermore, in this example, the data of the spectrum of the reflected light of each of a plurality of wafers that were pre-processed using isometric mapping is mapped based on the similarity. Figure 13 A graph showing the result of correlating the maximum value of the variance of the distance between each data obtained using isometric mapping with the initial thickness of the mask layer of each wafer is shown.

[0149] Figure 13 is a graph showing an example of correlating the maximum value of the variance of the distance between each data obtained using isometric mapping with the initial thickness of the mask layer of each wafer. Figure 1 A further other modification of the plasma processing apparatus according to the embodiment shown in Figure 2 is a graph showing an example of correlating the maximum value of the variance of the distance between each data obtained using isometric mapping with the initial thickness of the mask layer of each wafer. It can be confirmed that the first component based on isometric mapping shows a high correlation with the mask film thickness.

[0150] In this example, furthermore, based on the obtained first component values, the wafers are sequenced in correspondence with the similarity between each of the wafers, and based on the result thereof, a database of the spectrum of the reflected light is selected as the data of the waveform pattern database 122. Using these plurality of data, the amount of processing at a plurality of sampling times in the processing of an arbitrary wafer is detected, and using the result thereof, end point determination is implemented, and as a result of this end point determination, the wafer is selected as the wafer to be processed next. Figure 6 As in (b) of the same, the processed residual film thickness is 130 nm in proximity to the target in all wafers, and the error is also 0.5 nm or less. Therefore, in this example as well, it is apparent that even in a case where the amount of deviation in the characteristics of the film structure is unknown, correct detection of the amount of processing can be performed, and high-precision end point determination can be achieved.

[0151] In the above-described embodiment or modification, as in Figure 6 or Figure 8As shown, the plurality of wafers are numbered sequentially in accordance with the magnitude of the value of the parameter indicating the degree of similarity between the data on the reflected light from the wafers obtained in the processing of each of the plurality of wafers that have been processed in advance, and the data on the reflected light from the wafer corresponding to the parameter value that substantially equally divides the maximum value and the minimum value among the differences (errors) of the respective values of the parameter that vary in accordance with the number of the plurality of wafers and the value of the sum of the errors, or the data on the reflected light from the wafer corresponding to the parameter value that substantially equally divides the difference between the maximum value and the minimum value, are used in the detection of the amount of processing at a plurality of sampling times in the processing of an arbitrary wafer.

[0152] In the case where such data is selected, from among the data involved in the plurality of wafers that have been processed in advance described above, the data on the reflected light corresponding to a certain wafer is selected as the data used in the detection of the amount of processing, the range of wafers in which the amount of processing can be detected with an error within a given allowable range can be used to select the data of the waveform pattern database 122. An example of selecting such data to detect the amount of processing is described below.

[0153] In this example, before the data of the waveform pattern database 122 is selected, the intensity data of the light of the reflected light of a plurality of wavelengths obtained in the processing of a plurality of wafers that are candidates for the database is calculated as an estimated film thickness by taking the data corresponding to an arbitrary one of the wafers as reference data and taking the remaining other wafers as objects of detection, using a simulation or the like, for a given amount of processing in the processing (for example, a specific remaining film thickness at the end point or the like). Furthermore, the error of the actual value of the specific remaining film thickness and the estimated film thickness is calculated for each of the other wafers that are objects of detection.

[0154] The calculation of the error described above is repeated with each of the plurality of wafers as the wafer for reference data and the other wafers as the wafers that are objects of detection, and a table or matrix is obtained in which the value of the error of the given remaining film thickness of each of the other wafers that is detected with each of the plurality of wafers as reference is taken as a component, for each of the plurality of wafers. Figure 14 An example of such a table is shown.

[0155] Figure 14 is an example of a table of data in which the value of the error of the given remaining film thickness of each of the other wafers that is detected with each of a plurality of data on the intensity of the light of the reflected light from the film structure of the wafer surface is taken as a component, in which the intensity of the light of the reflected light from the film structure of the wafer surface is Figure 1 Yet another other modification of the plasma processing apparatus involved in the embodiment shown uses a plurality of wafers having a film structure as shown in Figure 2 to perform etching processing, and the data on the reflected light obtained in the processing. In this example, the data of the waveform pattern database 122 is selected using the table shown in Figure 14

[0156] ​Furthermore, in this example, in Figure 14 The table shown indicates that if the estimated film thickness error is within a given tolerance range (e.g., less than 1 nm), it is determined that the remaining film thickness can be detected. That is, by comparing the estimated film thickness error of the other (n-1) wafers detected using data from reflected light corresponding to any one of the n wafers each labeled with a number (database wafer number), with the upper limit of the tolerance range of 1 nm, a combination of wafers can be obtained among the other (n-1) wafers that can be used to detect the remaining film thickness using the data (comparison data) of that one wafer. In other words, a combination of database wafers that can perform film thickness detection within the aforementioned tolerance range can be selected from the n wafers for which film thickness estimation is required.

[0157] There are multiple combinations of such database chips, and these combinations become Figure 15 The data table shown. Figure 15 It means from Figure 14 The data table involved in other variations of the embodiments of the present invention shown is a table of combinations of database wafers capable of detecting the remaining film thickness of n wafers. Here, Figure 15 The table lists the combinations starting with the ones using the fewest database chips. For example, if you choose the combination with the fewest database chips... Figure 15 The first combination selects the reflected light data corresponding to the database chip as the data for waveform pattern database 122.

[0158] As a result of the endpoint determination implemented by using data from the waveform pattern database 122 selected in this way to detect the amount of processing in the processing of any chip, and Figure 6 Similarly, in (b), the processed film thickness is close to the target 130 nm across all wafers, with an error of less than 0.5 nm. Therefore, in this example, it is evident that even when the deviation in the film structure characteristics is unknown, accurate processing quantity detection can be performed, enabling high-precision endpoint determination.

[0159] Next, we will explain how to use it. Figure 1 Example (b) of the fitted database calculator 120 detecting the amount of processing at multiple sampling moments during wafer processing and determining the endpoint of processing. In this example, conditions other than those used by the fitted database calculator 120 are also set as described above. Figures 1 to 15 The embodiments or variations shown have the same structure.

[0160] The database is selected in the same manner as in Embodiment 1, and the data of the selected wafer is determined as the database of the waveform pattern database 122. The determined plurality of databases are used to determine the end point judgment based on the processing amount monitoring, but in this embodiment, the database of the waveform pattern database 122 is calculated using the fitting database calculator 120 of (b) of the above-described (a), and is used in the film thickness estimation. The specific process is described below. Figure 1

[0161] The spectrum at the time of measurement from the wafer to be estimated for the film thickness is obtained, and the data D2 obtained by signal processing of the spectrum in the digital signal processing section 100 is supplied to the fitting database calculator 120. In the fitting database calculator 120, a plurality of databases are also supplied from the waveform pattern database 122, and for example, a mixed database DBnm(a) (= a x DBn + (1 - a) x DBm) in which each of the two databases DBn and DBm is mixed at an arbitrary mixing ratio a is generated. In the fitting database calculator 120, the data D2 and the mixed database DBnm(a) are compared to determine the combination of the database numbers n and m and the mixing ratio a for which the error is the smallest with respect to the data D2.

[0162] The determined mixed database DBnm(a) is supplied to the waveform pattern database 122, and in the waveform comparator 102, the database of the waveform pattern database 122 including the mixed database DBnm(a) is used to determine the film thickness at the current time. In addition, the mixed database DBnm(a) used at the current time is supplied to the optimum database determiner 124 together with the combination of the database numbers n and m and the mixing ratio a.

[0163] The determination of the optimum database in the optimum database determiner 124 can be the database in which the error with respect to the measured spectrum is the smallest, in the same manner as in Embodiment 1. In addition, in this embodiment, since the mixed database having a spectrum close to the measured spectrum at each time is generated, the mixed database at each time can be determined as the optimum database.

[0164] The film thickness at each time is determined from the optimum film thickness / depth determiner 106 based on the determined optimum database. In this embodiment, as a result of the film thickness estimation in which the mixed database at each time is implemented as the optimum database, the film thickness of the processed wafer is close to the target 130 nm in all wafers, and the film thickness estimation error is reduced compared to Embodiment 1. Figure 6

[0165] ​​As in the present embodiment, by generating a database of spectra close to the wafer for which the film thickness is to be estimated by the fitting database generator 120, it is possible to perform film thickness estimation with higher accuracy than when the data of the wafer that has been processed in advance is set as the database. According to the above, in the throughput monitoring using the plurality of databases of the present embodiment, it is also apparent that, in the case where the structure bias is unknown, it is possible to perform correct measurement of the throughput, and it is possible to achieve high-accuracy end-point determination.

[0166] In the example described above, the technique of detecting the throughput in the processing of the wafer in the case where the information on the characteristics of the film structure of the wafer is unknown was described. Next, the case where the information on the characteristics of the film structure other than the processing target film of the wafer is obtained before the start of the processing of the wafer. In the present example, with respect to the conditions other than this, the same structure as in Embodiment 1 is assumed.

[0167] In the present example, in the case where, for example, the initial film thickness of the mask layer of each wafer is obtained before the start of the processing as the information on the characteristics of the film structure other than the processing target film, the data stored in the waveform pattern database 122 and used in the detection of the throughput is decided using the information on the initial film thickness of the mask layer. First, the wafers are sequenced in correspondence with the magnitude of the value of the initial film thickness of the mask layer of each wafer, and the wafers are numbered in order from the largest to the smallest film thickness, and the data of the wafer corresponding to each of the values substantially equally spaced between the maximum and minimum values of the initial film thickness of the mask layer is decided. Figure 6 The wafer number on the horizontal axis in (a) of the above. As in the above embodiment, the data of the reflected light of the wafer of the number corresponding to the maximum, minimum, and each value substantially equally spaced between them of the initial film thickness of the mask layer is selected as the data of the waveform pattern database 122.

[0168] As a result of the processing of the target wafer using the plurality of selected data, the detection of the throughput in the processing, and the implementation of the end-point determination based on the result thereof, the same as in (b) of the above Figure 6 As in the above embodiment, the film thickness after the processing is near the target 130 nm in all of the wafers, and the error is also 0.5 nm or less. Therefore, in the throughput monitoring using the plurality of databases of the present embodiment, it is also apparent that, in the case where the structure bias is unknown, it is possible to perform correct measurement of the throughput, and it is possible to achieve high-accuracy end-point determination.

[0169] As in the present embodiment, in the case where the information on the characteristics of the film structure of the wafer is unknown, it is possible to perform the detection of the throughput in the processing of the wafer with high accuracy. Figure 19 An example of the structure of a processing apparatus and a monitoring system of information between processing apparatuses that manages the case where the information on the wafer obtained before the processing, particularly the information on the characteristics of the film structure, is shown. Figure 19 is a schematic representation of the monitoring Figure 1 is a schematic block diagram showing the structure of the system of the plasma processing apparatus to which the embodiment shown in

[0170] In the example shown in the figure, characteristics such as the initial film thickness and film structure dimensions of the mask before processing, including the wafer in processing state 1, are detected by a detector (not shown) in processing device A. Data representing the detection results is supplied to processing device A, and the wafer is processed in processing device A using this data. Information on the characteristics of the film structure of the wafer in processing state 1, and information on the processing status including the amount of film layer processed, are supplied to a monitoring system, such as monitoring device A or B, which can be communicatively connected to each processing device. This information can also be used in other processing devices, such as processing devices B, C, and D, which can be communicatively connected to these monitoring devices.

[0171] Furthermore, information such as the processing conditions of each wafer in each processing unit can also be used in subsequent processing units via a monitoring system. Therefore, in each processing unit, the same database selection as in this example can be performed based on information from previous processing and measurements. By implementing database selection in each processing unit that utilizes information about the wafer's processing status, high-precision processing can be achieved.

[0172] Next, we will explain an example of selecting data for detecting reflected light in the quantity of processing when the position and range of reflected light on the wafer change during the detection process. Conditions other than this point will be assumed to be... Figures 1 to 15 The embodiments or variations shown have the same structure.

[0173] Figure 16 It is a schematic representation of the detection. Figure 1 The plasma processing apparatus according to another variation of the illustrated embodiment is for having Figure 2 The figure shows a top view of the wafer surface where reflected light from the wafers is located during etching of multiple wafers with the film structure shown. In particular, this figure illustrates the location of reflected light on a wafer detected by two different plasma processing devices on a hypothetical wafer.

[0174] As in this example, when the location, wavelength, and intensity of the reflected light differ in each of the multiple plasma processing devices, the initial film thickness of the mask being detected may vary. Figure 2 As shown in (b), this deviation causes a change in the correlation between the film thickness of the etched film in each device and the spectrum of the reflected light from it. Furthermore, due to the differences in the wavelength and intensity range of the reflected light spectrum, the range of detectable light varies. Figure 2 The proportions of the pattern shown in (a) and other patterns vary, so the relationship between the film thickness of the etched object and the spectrum of reflected light may vary depending on the wafer.

[0175] The inventors possessed and Figure 1The same as the embodiment, the data of the spectrum of the reflected light of each of the plurality of wafers processed by each of the plasma processing apparatuses is extracted. Using the extracted data, the error amount of the spectrum of the wafer processed by each of the plasma processing apparatuses is calculated in the same manner as the embodiment, and the results of the error amount plotted against each number for identification of the plasma processing apparatus are shown. Figure 17 The results of the error amount plotted against each number for identification of the plasma processing apparatus are shown.

[0176] Figure 17 A graph showing the sum of the error of a given residual film thickness of other wafers for each of which a plurality of data of the intensity of light of the reflected light from the film structure on the wafer surface is detected is shown for each of the plurality of plasma processing apparatuses, wherein the intensity of light of the reflected light from the film structure on the wafer surface is Figure 1 Another other modification of the embodiment shown involves a plurality of plasma processing apparatuses that have Figure 2 The plurality of wafers each having the film structure shown are subjected to etching processing, and the data of the spectrum of the reflected light obtained in the processing is known to differ among the plasma processing apparatuses. In this case, the data of the reflected light of the wafer of the plasma processing apparatus corresponding to the maximum value and the minimum value of the error amount and the value therebetween are selected as the data stored in the waveform pattern database 122 and used in the detection of the processing amount.

[0177] As the processing amount at the plurality of sampling times in the processing of an arbitrary wafer using the selected plurality of data, the results of the end point determination implemented using the detection results are used, and the same as the embodiment Figure 6 As in (b) of the embodiment, the film thickness after the processing is near the target 130 nm in all the wafers, and the error is also 0.5 nm or less. Therefore, it is also apparent that in the case where the relationship between the film thickness and the spectrum differs among the apparatuses, correct measurement of the processing amount can be performed in this case, and high-precision end point determination can be achieved.

[0178] Next, an example of selecting the data of the reflected light used in the detection of the processing amount in the case where the selection ratio of the material of the film layer to be processed and the other film layers of the film structure on the wafer is changed is described. As for the conditions other than this, the same structure as the embodiment shown or the modification is assumed. Figures 1 to 17

[0179] In this case, the film structure including the film layer to be etched on the wafer is also the same as the embodiment shown. Figure 2 ​Similar to the film structure shown in (a), the initial film thickness of the mask layer is substantially the same across multiple wafers. However, in this example, the selection ratio of the mask layer to the film layer to be processed varies during the processing of each wafer. Therefore, a situation arises where, even when the remaining film thickness of the film layer to be processed is the same in each wafer, the amount of mask layer removed differs among these wafers, resulting in different remaining film thicknesses for the mask on each wafer. In this case, even if the remaining film thickness of the film layer to be processed is the same in multiple wafers, the spectral data of the reflected light from the surfaces of these wafers are different, creating the same problem as the variation in the initial film thickness of the mask layer in the above embodiment.

[0180] In this example, with Figure 1 Similarly, in the implementation method, the remaining film thickness correspondence is established, the film thickness range is determined, and the spectral data of reflected light obtained from the processing of each of the multiple pre-processed wafers are interpolated to extract spectral data of the same remaining film thickness from the data of the processing of each of the multiple wafers. As the average error between the extracted spectral data and all wafers, the result of establishing a correspondence between the calculated error and the film thickness of the mask of each wafer is compared with... Figure 5 Similarly, in (c), the error and the remaining film thickness of the mask are highly correlated. Based on the error amount, the wafers are sorted and serialized. Based on the sorting result, data of the spectrum of reflected light corresponding to multiple wafers are selected as data for the waveform pattern database 122, similar to the implementation method.

[0181] As a result of endpoint determination implemented using multiple selected data points, and in accordance with the implementation method... Figure 6 Similarly, in (b), the processed film thickness is close to the target 130 nm across all wafers, with an error of less than 0.5 nm. Therefore, it is evident that, in this example, accurate measurement of the processing amount and high-precision endpoint determination can be achieved, even when the relationship between film thickness and spectrum differs between devices.

[0182] Next, the explanation is in Figure 1 (b) shows an example of using the time shift of the detected remaining film thickness in the optimal database determination of the optimal database determiner 124. For conditions other than this, the same structure as in Implementation 1 applies.

[0183] exist Figure 1 The film thickness / depth calculation unit 30 uses the database to detect the time variation of the remaining film thickness at each sampling moment in the etching process, for example, as shown in the example. Figure 18 Data 1 and 2. Figure 18 It means Figure 1 The plasma processing apparatus according to another variation of the illustrated embodiment is for having Figure 2A chart of an example of a change in the residual film thickness of the film layer of the processing target on the wafer with the passage of time accompanying the processing of the wafer using the reflected light from the wafer is shown. In this example, the correlation between the detected residual film thickness and the time in the processing is shown based on the first linear correlation of each data, and the residual film thickness is estimated.

[0184] For example, at each sampling time in the processing, the correlation coefficient is calculated from the values of the residual film detected at a plurality of times in the past up to that time and the times in the past, and the data with a large correlation coefficient is selected as the data of the optimum database. For example, when the correlation coefficient is obtained Figure 18 In the case where the mutual relationship can be substantially represented by a straight line as shown in the two data 1 and 2, the absolute value of the correlation coefficient of the data 1 is 0.99, but the absolute value of the correlation coefficient of the data 2 is 0.95. Therefore, in this example, the data 1 with a large correlation coefficient is selected as the data of the optimum database.

[0185] The amount of processing is detected at a plurality of sampling times in the processing using the data of the spectrum of the reflected light selected as the data of the optimum database, and the result is used to implement the end point determination. The processing of the embodiment Figure 6 In (b) of the same, the film thickness after the processing is near the target 130 nm in all the wafers, and the error is also 0.5 nm or less. Therefore, in this example, it is also apparent that the correct measurement of the amount of processing can be performed even in the case where the construction deviation is unknown, and the high-precision end point determination can be achieved.

[0186] Explanation of reference numerals

[0187] 10... processing chamber

[0188] 12... plasma

[0189] 14... sample stage

[0190] 16... processing target

[0191] 18... light source section

[0192] 20... introduction lens

[0193] 22... irradiation light

[0194] 24... reflected light

[0195] 26... detection lens

[0196] 28... detection section

[0197] 30... film thickness / depth calculation section

[0198] 40... control section

[0199] 50 optical system

[0200] 60 database selection section

[0201] 100 digital signal processing section

[0202] 102 waveform comparator

[0203] 104 film thickness / depth storage section

[0204] 106 optimum film thickness / depth decider

[0205] 120 fitting database calculator

[0206] 122 waveform pattern database

[0207] 124 optimum database decider

[0208] D1 time series data supplied from the detection section

[0209] D2 time series data supplied from the digital signal processing section

[0210] D3 film thickness / depth data supplied from the waveform comparator

[0211] D4 film thickness / depth data supplied from the film thickness / depth storage section

Claims

1. A plasma processing apparatus for processing a wafer of a processing object disposed within a processing chamber inside a vacuum container using plasma formed in the processing chamber, the plasma processing apparatus being characterized by comprising: A light receiver that receives light of multiple wavelengths from the surface of the wafer at given multiple moments during the processing of the wafer being processed; and A detector uses the result of comparing data representing the intensity of light at the multiple wavelengths received with comparative data representing the intensity of light at the multiple wavelengths obtained in advance to detect the amount of processing in the wafer being processed. The detector quantifies the similarity between the wafers based on data representing the intensity of multiple wavelengths of light from the surface of each wafer, obtained in advance during the respective processing of the multiple wafers. It then selects at least one data point based on the quantified similarity as the comparison data and compares this comparison data with data representing the intensity of the multiple wavelengths of light obtained during the processing of the wafer to be processed, thereby detecting the amount of processing. The detector quantifies the similarity by taking the difference between the spectral values ​​of each film thickness at each wavelength and the average value of the spectrum as an error, based on multiple data representing the intensity of light at the multiple wavelengths obtained in the respective processing of the multiple wafers. The detector also uses a database of patterns of the spectra of wafers that are the wafers with the largest and smallest sums of the errors, and wafers whose sums of the errors that make the wafers with the largest sums of the errors and the wafers with the smallest sums of the errors, as well as the difference between the sums of the errors that make the wafers with the largest sums of the errors and the sums of the errors that make the wafers with the smallest sums of the errors, as the database, and selects data as the comparison data based on the database.

2. The plasma treatment apparatus according to claim 1, characterized in that, The plasma processing apparatus uses the difference between the intensity of the light and the average of the light intensities, or the square of the difference, which represents multiple data points indicating the intensity of the multiple wavelengths of light obtained in advance in the respective processing of the multiple wafers, as an index representing the similarity.

3. The plasma treatment apparatus according to claim 1, characterized in that, The plasma processing apparatus uses principal component analysis to obtain principal component values ​​from multiple data representing the intensity of multiple wavelengths of light obtained in the respective processing of the multiple wafers, and uses these values ​​as an index representing the similarity.

4. The plasma treatment apparatus according to claim 1, characterized in that, The plasma processing apparatus uses the value of the dimensionality-reduced component, obtained as a result of performing a dimensionality-reduction technique on multiple data representing the intensity of multiple wavelengths of light obtained in the respective processing of the multiple wafers, as an index representing the similarity.

5. The plasma treatment apparatus according to claim 1, characterized in that, The at least one data point is selected from a plurality of data representing the intensity of light at a plurality of wavelengths obtained in advance in the respective processes of the plurality of wafers, wherein the similarity between the at least one data point and the other data is within a predetermined given allowable range.

6. A plasma processing method for processing a wafer of a processing object disposed in a processing chamber inside a vacuum container using plasma formed in the processing chamber, the plasma processing method being characterized by having: The measurement process involves receiving light of multiple wavelengths from the surface of the wafer at given multiple moments during the processing of the wafer being processed; and The detection process uses the result of comparing data representing the intensity of light at the multiple wavelengths received with comparative data representing the intensity of light at the multiple wavelengths obtained in advance to detect the amount of processing in the wafer being processed. In the detection process, based on data representing the intensity of light at multiple wavelengths from the surface of each wafer obtained in the respective processing of multiple wafers, the similarity between the wafers is quantified. At least one data point selected based on the quantified similarity is used as the comparison data. This comparison data is compared with data representing the intensity of light at multiple wavelengths obtained in the processing of the wafer to be processed to detect the amount of processing. The similarity is quantified by taking the difference between the spectral values ​​of each film thickness at each wavelength and the average value of the spectrum as an error, based on multiple data representing the intensity of light at each wavelength obtained in the respective processing of the multiple wafers. A database is created of the spectral patterns of wafers whose sum of errors is the largest and smallest wafers, and whose values ​​are substantially equal divisions between the sum of errors of the wafers that make the sum of errors of the largest wafer and the sum of errors of the wafers that make the sum of errors of the smallest wafer. Data is then selected as the comparison data based on the database.

7. The plasma treatment method according to claim 6, characterized in that, The difference between the intensity of the light and the average of the light intensities, or the square of the difference, which represents multiple data points indicating the intensity of the multiple wavelengths of light obtained in advance in the respective processes of the multiple wafers, is used as an indicator of the similarity.

8. The plasma treatment method according to claim 6, characterized in that, The principal component values ​​obtained by performing principal component analysis on multiple data representing the intensity of light at multiple wavelengths obtained in the respective processes of the multiple wafers are used as an index representing the similarity.

9. The plasma treatment method according to claim 6, characterized in that, The value of the dimension-reduced component, obtained as a result of performing a dimension reduction technique on multiple data representing the intensity of light at multiple wavelengths obtained in the respective processes of the multiple wafers, is used as an index representing the similarity.

10. The plasma treatment method according to claim 6, characterized in that, The at least one data point is selected from a plurality of data representing the intensity of light at a plurality of wavelengths obtained in advance in the respective processes of the plurality of wafers, wherein the similarity between the at least one data point and the other data is within a predetermined given allowable range.

Images