Method and grinding device for inspecting spectrum of reflected light from workpiece

Abstract

The invention provides a method and a polishing apparatus for inspecting the spectrum of reflected light from a workpiece during polishing of the workpiece such as a wafer. In this method, light is irradiated from an optical sensor head to a plurality of film thickness measurement points on a workpiece (W) for each time interval during polishing of the workpiece (W), a plurality of measurement spectrums of reflected light from the plurality of film thickness measurement points are generated, a plurality of feature quantities of the plurality of measurement spectrums are calculated, and the plurality of feature quantities are calculated. A plurality of times at which a plurality of measurement spectrums have been generated and a plurality of measurement data points (MD) specified on the basis of a plurality of feature quantities are plotted on a coordinate system, thereby performing data mapping, and determining a correction index value indicating the number of the plurality of measurement data points (MD) present within a threshold range (TR) specified on the coordinate system for each time interval. And determining whether the correction index value is within the correction management range.

Description

TECHNICAL FIELD

[0001] The present application relates to a technique of measuring a film thickness of a workpiece based on a spectrum of reflected light from the workpiece such as a wafer, a substrate, a panel, and the like, and particularly relates to a technique of detecting an abnormal spectrum of reflected light from the workpiece. BACKGROUND

[0002] Various processes including a process of polishing an insulating film such as SiO2 and a process of polishing a metal film such as copper and tungsten are included in a manufacturing process of a semiconductor device. Polishing of a wafer is performed using a polishing apparatus. The polishing apparatus generally has a polishing table that supports a polishing pad, a polishing head that presses a wafer against the polishing pad, and a slurry supply nozzle that supplies slurry to the polishing pad. While the polishing table is rotated, slurry is supplied to the polishing pad on the polishing table, and the polishing head presses the wafer against the polishing pad. The wafer is brought into sliding contact with the polishing pad in the presence of the slurry. The surface of the wafer is planarized by a combination of chemical action of the slurry and mechanical action of the polishing pad and abrasive grains contained in the slurry.

[0003] Polishing of a wafer is ended when the thickness of a film (an insulating film, a metal film, a silicon layer, or the like) constituting the surface thereof reaches a prescribed target value. The polishing apparatus generally has an optical film thickness measurement system in order to measure the thickness of a non-metal film such as an insulating film, a silicon layer, or the like. The optical film thickness measurement system is configured to direct light of a light source to the surface of the wafer, measure the intensity of reflected light from the wafer using a light splitter, analyze the spectrum of the reflected light, and thereby measure the film thickness of the wafer.

[0004] PRIOR ART DOCUMENTS

[0005] PATENT DOCUMENTS

[0006] Patent Document 1: Japanese Patent Application Publication No. 2022-156879

[0007] Due to failure or deterioration of a light source, an optical system, or the like over the years, the intensity of reflected light from a wafer changes, and as a result, an abnormal spectrum is sometimes acquired. In other examples, when a wafer of a type different from a target wafer that is supposed to be polished is sent to the polishing apparatus, a spectrum of reflected light from the wafer is acquired as an abnormal spectrum. Such an abnormal spectrum becomes a cause of an erroneous film thickness measurement value. In addition, if the polishing operation of the wafer is controlled based on a film thickness measurement value obtained from the abnormal spectrum, a desired polishing result cannot be obtained. SUMMARY

[0008] Therefore, the present application provides a technique of checking a spectrum of reflected light from a workpiece such as a wafer in polishing of the workpiece.

[0009] In one embodiment, a method of inspecting a spectrum of reflected light from a workpiece is provided, wherein the workpiece is polished while being pressed against a polishing pad on a polishing table that is rotated, at each time interval in the polishing of the workpiece, light is irradiated from an optical sensor head to a plurality of film thickness measurement points on the workpiece, a plurality of measurement spectra of reflected light from the plurality of film thickness measurement points is generated, a plurality of characteristic amounts of the plurality of measurement spectra is calculated, a plurality of times at which the plurality of measurement spectra is generated and a plurality of measurement data points determined from the plurality of characteristic amounts of the plurality of measurement spectra are plotted on a coordinate system, thereby performing data mapping, a correction index value that indicates a number of the plurality of measurement data points present within a threshold value range determined on the coordinate system at each time interval is decided, and whether the correction index value is within a correction management range is determined.

[0010] In one embodiment, when the correction index value is within the correction management range, the plurality of measurement data points is moved on the coordinate system until the plurality of measurement data points converges within the threshold value range, thereby correcting the plurality of characteristic amounts.

[0011] In one embodiment, when the correction index value is less than a lower limit of the correction management range, an alarm signal is generated.

[0012] In one embodiment, further comprising the steps of: polishing a reference workpiece while being pressed against the polishing pad on the polishing table that is rotated, at each time interval in the polishing of the reference workpiece, light is irradiated from the optical sensor head to a plurality of film thickness measurement points on the reference workpiece, a plurality of reference spectra of reflected light from the plurality of film thickness measurement points is generated, a plurality of characteristic amounts of the plurality of reference spectra is calculated, a plurality of times at which the plurality of reference spectra is generated and a plurality of reference data points determined from the plurality of characteristic amounts of the plurality of reference spectra are plotted on the coordinate system, thereby performing data mapping, and the threshold value range at each time interval in the polishing of the reference workpiece is made based on the plurality of reference data points on the coordinate system.

[0013] In one embodiment, the threshold value range is a range within a prescribed Mahalanobis distance from a reference point of the plurality of reference data points.

[0014] In one embodiment, when the correction index value is greater than an upper limit of the correction management range, the plurality of reference data points is updated by adding the plurality of measurement data points to the plurality of reference data points, and the threshold value range is updated based on the plurality of reference data points after the update.

[0015] In one embodiment, the method further includes the step of selecting either a first set of a plurality of threshold ranges or a second set of a plurality of threshold ranges according to the progress of the plurality of measured data points with the polishing time of the workpiece, the threshold range being one of the plurality of threshold ranges of the selected one of the first set and the second set, the first set of the plurality of threshold ranges being predetermined from a plurality of reference data points obtained in polishing of a first reference workpiece, the second set of the plurality of threshold ranges being predetermined from a plurality of reference data points obtained in polishing of a second reference workpiece having a different surface configuration from the first reference workpiece.

[0016] In one embodiment, the reference workpiece is a first reference workpiece and a second reference workpiece, and the method further includes the step of creating a plurality of threshold ranges corresponding to different time intervals by combining a first threshold range created from a plurality of reference data points obtained in polishing of the first reference workpiece with a second threshold range created from a plurality of reference data points obtained in polishing of the second reference workpiece.

[0017] In one embodiment, the optical sensor head is a first optical sensor head and a second optical sensor head disposed at different positions in the polishing table, and the method further includes the step of creating a plurality of threshold ranges corresponding to different time intervals by combining a first threshold range created from a plurality of reference data points obtained in polishing of the reference workpiece by light irradiation from the first optical sensor head with a second threshold range created from a plurality of reference data points obtained in polishing of the reference workpiece by light irradiation from the second optical sensor head.

[0018] In one embodiment, the plurality of characteristic amounts each include at least a k-th principal component obtained by performing principal component analysis on a data set including a plurality of intensities of the reflected light at a plurality of wavelengths of each measured spectrum, where k is a natural number.

[0019] In one embodiment, in a plurality of consecutive time intervals in polishing of the workpiece, when the correction index value is smaller than the lower limit of the correction management range, a plurality of new threshold ranges corresponding to the plurality of consecutive time intervals are created from a plurality of measured data points obtained in the plurality of consecutive time intervals.

[0020] In one embodiment, there is provided a polishing apparatus, which is a polishing apparatus of a workpiece, including: a polishing table; a table motor that rotates the polishing table; a polishing head that polishes the workpiece by pressing the workpiece against a polishing pad on the polishing table; an optical sensor head that irradiates light to a plurality of film thickness measurement points on the workpiece at each time interval during polishing of the workpiece; and a processing system that generates a plurality of measurement spectra of reflected light from the plurality of film thickness measurement points, the processing system configured to calculate a plurality of feature amounts of the plurality of measurement spectra, and plot a plurality of times at which the plurality of measurement spectra are generated and a plurality of measurement data points determined from the plurality of feature amounts of the plurality of measurement spectra on a coordinate system, thereby performing data mapping, and determine a correction index value that indicates a number of the plurality of measurement data points present within a threshold value range determined on the coordinate system at each time interval, and determine whether the correction index value is within a correction management range.

[0021] In one embodiment, the processing system is configured to correct the plurality of feature amounts by moving the plurality of measurement data points on the coordinate system until the plurality of measurement data points converge within the threshold value range, when the correction index value is within the correction management range.

[0022] In one embodiment, the processing system is configured to generate an alarm signal when the correction index value is less than a lower limit of the correction management range.

[0023] In one embodiment, the processing system is configured to select either a first group of a plurality of threshold value ranges or a second group of a plurality of threshold value ranges according to a progression of the plurality of measurement data points with a polishing time of the workpiece, the threshold value range being one of a plurality of threshold value ranges of the selected one of the first group and the second group, the first group of a plurality of threshold value ranges being determined in advance from reference data points obtained in polishing of a first reference workpiece, the second group of a plurality of threshold value ranges being determined in advance from reference data points obtained in polishing of a second reference workpiece having a different surface configuration from the first reference workpiece.

[0024] In one embodiment, the processing system is configured to perform principal component analysis on a data set including a plurality of intensities of the reflected light at a plurality of wavelengths of each measurement spectrum, and determine a feature amount including at least a kth principal component obtained from the principal component analysis, where k is a natural number.

[0025] In one embodiment, the processing system is configured to, when the correction index value is smaller than the lower limit of the correction management range, create a plurality of new threshold value ranges corresponding to a plurality of continuous time intervals in the polishing of the workpiece, based on a plurality of measurement data points acquired in the plurality of continuous time intervals.

[0026] In one embodiment, the processing system is configured to determine the threshold value range as a range within a prescribed Mahalanobis distance, based on a reference point of a plurality of reference data points obtained from the polishing of a reference workpiece.

[0027] In one embodiment, the processing system is configured to, when the correction index value is larger than the upper limit of the correction management range, update the plurality of reference data points by adding the plurality of measurement data points to the plurality of reference data points, and update the threshold value range based on the plurality of updated reference data points.

[0028] Effects of the Invention

[0029] By comparing the measurement data points with the threshold value range for each time interval within the polishing time of the workpiece, the processing system can determine whether the measurement spectrum is normal based on the comparison result. Since the threshold value range is set for each time interval, the polishing of the workpiece can be stopped at a point in time when the measurement data points greatly deviate from the threshold value range. As a result, over-polishing of the workpiece can be prevented. In addition, damage to a subsequent workpiece due to an erroneous polishing process can be prevented. The workpiece can be re-polished under different polishing conditions, or can be taken out of the polishing device without being re-polished. As a result, the response time after abnormality detection is shortened. If the abnormality of the measurement spectrum is due to the optical film thickness measurement device, repair of the optical film thickness measurement device can be performed. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is a schematic view showing one embodiment of a polishing device.

[0031] Figure 2 is a cross-sectional view showing a detailed structure of an optical film thickness measurement device.

[0032] Figure 3 is a schematic view showing an example of a measurement spectrum generated from light intensity measurement data.

[0033] Figure 4 is a view showing an example of a film thickness measurement point of light on a workpiece.

[0034] Figure 5 is a view showing one embodiment of data mapping.

[0035] Figure 6is a graph indicating a plurality of measurement data points MD N acquired at the time of the Nth rotation of the polishing table.

[0036] Figure 7 is a graph indicating a plurality of measurement data points MD N+1 acquired at the time of the N+lth rotation of the polishing table.

[0037] Figure 8 is a graph indicating a plurality of measurement data points MD N+2 acquired at the time of the N+2th rotation of the polishing table.

[0038] Figure 9 is a graph indicating one embodiment in which the plurality of measurement data points MD N+2 are moved until the plurality of measurement data points MD N+2 converge within the threshold range TR N+2 .

[0039] Figure 10 is a graph indicating one example in which all or a large portion of the plurality of measurement data points MD N+3 exist outside the threshold range TR N+3 .

[0040] Figure 11 is a flowchart of the embodiment described with reference to Figures 1 to 10 .

[0041] Figure 12 is a graph indicating one embodiment in which a first group and a second group of a plurality of threshold ranges are prepared.

[0042] Figure 13 is a graph indicating one example of a change over time of measurement data points plotted on a coordinate system in polishing of a workpiece.

[0043] Figure 14 is a flowchart of the embodiment described with reference to Figure 12 and Figure 13 .

[0044] Figure 15 is a graph indicating one embodiment in which a new threshold range is created in polishing of a workpiece.

[0045] Figure 16 is a graph indicating one embodiment in which a first threshold range and a second threshold range are created based on reference data points obtained in polishing of a first reference workpiece and a second reference workpiece.

[0046] Figure 17 is a plan view indicating one example of a positional relationship between an optical sensor head and a polishing head.

[0047] Figure 18is a diagram showing one embodiment of making a plurality of threshold value ranges corresponding to different time intervals by combining the first threshold value range and the second threshold value range.

[0048] Figure 19 is a plan view showing one embodiment of a polishing apparatus having a first optical sensor head and a second optical sensor head.

[0049] Figure 20 is a diagram showing an example of a plurality of first threshold value ranges made from a plurality of reference data points obtained by light irradiation from the first optical sensor head and a plurality of second threshold value ranges made from a plurality of reference data points obtained by light irradiation from the second optical sensor head.

[0050] Figure 21 is a diagram showing a plurality of first threshold value ranges and a plurality of second threshold value ranges arranged along a time axis.

[0051] Figure 22 is a diagram showing a plurality of first threshold value ranges and a plurality of second threshold value ranges moved along a characteristic amount axis.

[0052] Symbol explanation

[0053] 1 Polishing head

[0054] 2 Polishing pad

[0055] 2a Polishing surface

[0056] 3 Polishing table

[0057] 5 Polishing liquid supply nozzle

[0058] 6 Table motor

[0059] 9 Motion control section

[0060] 10 Head shaft

[0061] 15 Polishing head rotating device

[0062] 20 Optical film thickness measuring device

[0063] 22 Light source

[0064] 25 Optical sensor head

[0065] 27 Light splitter

[0066] 30 Processing system

[0067] 31 Light projecting optical fiber cable

[0068] 32 Light receiving optical fiber cable

[0069] W Workpiece Detailed Implementation

[0070] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. Figure 1 This is a schematic diagram illustrating one embodiment of the grinding apparatus. (For example...) Figure 1 As shown, the grinding apparatus includes: a grinding table 3 supporting a grinding pad 2; a grinding head 1 pressing a workpiece W onto the grinding pad 2; a stage motor 6 rotating the grinding table 3; a grinding slurry supply nozzle 5 supplying grinding slurry such as slurry to the grinding pad 2; and an operation control unit 9 controlling the operation of the grinding apparatus. The upper surface of the grinding pad 2 forms a grinding surface 2a for grinding the workpiece W. The workpiece W has a film forming a wiring structure on its surface. Examples of the workpiece W include wafers, substrates, wiring substrates, and slab substrates used in the manufacture of semiconductor devices. In one example, the workpiece W is a product wafer with a multilayer or single-layer film formed on it.

[0071] The grinding head 1 is connected to the head shaft 10, and the head shaft 10 is connected to the grinding head rotating device 15. The grinding head rotating device 15 is configured to rotate the grinding head 1 and the head shaft 10 together in the direction indicated by the arrow. The structure of the grinding head rotating device 15 is not particularly limited; in one example, the grinding head rotating device 15 includes a motor, a belt, and pulleys. The grinding table 3 is connected to the table motor 6, and the table motor 6 is configured to rotate the grinding table 3 and the grinding pad 2 in the direction indicated by the arrow. The grinding head 1, the grinding head rotating device 15, and the table motor 6 are connected to the motion control unit 9.

[0072] Workpiece W is ground in the following manner. The table motor 6 and the grinding head rotation device 15 simultaneously move the grinding table 3 and the grinding head 1 towards... Figure 1 The grinding head rotates in the direction indicated by the arrow, supplying grinding fluid from the grinding fluid supply nozzle 5 to the grinding surface 2a of the grinding pad 2 on the grinding table 3. The workpiece W rotates through the grinding head 1 and is pressed against the grinding surface 2a of the grinding pad 2 while the grinding fluid is present on the grinding pad 2. The surface of the workpiece W is ground by the chemical action of the grinding fluid and the abrasive particles contained in the grinding fluid and / or the mechanical action of the grinding pad 2.

[0073] The motion control unit 9 includes a storage device 9a storing a program and a processing device 9b that performs calculations according to the commands contained in the program. The motion control unit 9 is composed of at least one computer. The storage device 9a includes a main storage device such as random access memory (RAM) and an auxiliary storage device such as a hard disk drive (HDD) or a solid-state drive (SSD). Examples of the processing device 9b include a CPU (central processing unit) and a GPU (graphics processing unit). However, the specific structure of the motion control unit 9 is not limited to these examples.

[0074] The polishing apparatus is provided with an optical film thickness measuring apparatus 20 that measures the film thickness of the workpiece W. The optical film thickness measuring apparatus 20 is provided with a light source 22 that emits light, an optical sensor head 25 that irradiates the light of the light source 22 toward the workpiece W and receives reflected light from the workpiece W, a light splitter 27 that is connected to the optical sensor head 25, and a processing system 30 that measures the film thickness of the workpiece W based on the spectrum of the reflected light from the workpiece W. The optical sensor head 25 is disposed in the polishing table 3 and rotates together with the polishing table 3.

[0075] The processing system 30 is provided with a storage device 30a in which a program is stored and an arithmetic device 30b that performs an operation according to a command included in the program. The processing system 30 is constituted by at least one computer. The storage device 30a is provided with a main storage device such as a random access memory (RAM) and a secondary storage device such as a hard disk drive (HDD) or a solid state drive (SSD). As examples of the arithmetic device 30b, a CPU (central processing unit) and a GPU (graphics processing unit) can be given. However, the specific structure of the processing system 30 is not limited to these examples.

[0076] The action control section 9 and the processing system 30 can each be constituted by a plurality of computers. For example, the action control section 9 and the processing system 30 can each be constituted by a combination of an edge server and a cloud server. In one embodiment, the action control section 9 and the processing system 30 can be constituted by one computer.

[0077] Figure 2 is a cross-sectional view that shows the detailed structure of the optical film thickness measuring apparatus 20. The optical film thickness measuring apparatus 20 is provided with an optical fiber cable 31 for light projection that is connected to the light source 22 and an optical fiber cable 32 for light reception that is connected to the light splitter 27. The tip end 31a of the optical fiber cable 31 for light projection and the tip end 32a of the optical fiber cable 32 for light reception constitute the optical sensor head 25. That is, the optical fiber cable 31 for light projection guides the light emitted by the light source 22 to the workpiece W on the polishing pad 2, and the optical fiber cable 32 for light reception receives the reflected light from the workpiece W and transmits it to the light splitter 27.

[0078] The light splitter 27 is connected to the processing system 30. The optical fiber cable 31 for light projection, the optical fiber cable 32 for light reception, the light source 22, and the light splitter 27 are installed to the polishing table 3 and integrally rotate together with the polishing table 3 and the polishing pad 2. The optical sensor head 25 constituted by the tip end 31a of the optical fiber cable 31 for light projection and the tip end 32a of the optical fiber cable 32 for light reception is disposed opposite to the surface of the workpiece W on the polishing pad 2.

[0079] The position of the optical sensor head 25 is the position at which the surface of the work W on the polishing pad 2 is crossed by the optical sensor head 25 every time the polishing table 3 and the polishing pad 2 rotate one revolution. The polishing pad 2 has a through-hole 2b above which the optical sensor head 25 is located. The optical sensor head 25 irradiates light to the work W through the through-hole 2b every time the polishing table 3 rotates one revolution, and receives reflected light from the work W through the through-hole 2b.

[0080] In one embodiment, a flow of pure water can also be formed in the through-hole 2b of the polishing pad 2 in order to prevent the polishing liquid and polishing slurry from coming into contact with the optical sensor head 25. Light from the optical sensor head 25 passes through the pure water and is directed to the work W, and reflected light from the work W passes through the pure water and is received by the optical sensor head 25. In other embodiments, a transparent window (not shown) can also be embedded in the through-hole 2b of the polishing pad 2. The transparent window is composed of a material that allows light to pass through, such as a transparent resin. In this case, light from the optical sensor head 25 passes through the transparent window and is directed to the work W, and reflected light from the work W passes through the transparent window and is received by the optical sensor head 25.

[0081] The light source 22 is a flash light source that repeatedly emits light at short time intervals. As an example of the light source 22, a xenon flash lamp can be given. The light source 22 is electrically connected to the operation control section 9, and emits light in response to a trigger signal transmitted from the operation control section 9. More specifically, the light source 22 receives a plurality of trigger signals and emits light a plurality of times during the period in which the optical sensor head 25 crosses the surface of the work W on the polishing pad 2. Thus, light is irradiated to a plurality of film thickness measurement points on the work W every time the polishing table 3 rotates one revolution.

[0082] Light emitted by the light source 22 is transmitted to the optical sensor head 25. That is, light is transmitted to the optical sensor head 25 through the light projecting optical fiber cable 31, and is radiated from the optical sensor head 25. The light enters the work W on the polishing pad 2 through the through-hole 2b of the polishing pad 2. The light reflected from the work W passes through the through-hole 2b of the polishing pad 2 again, and is received by the optical sensor head 25. The reflected light from the work W is transmitted to the light receiver 27 through the light receiving optical fiber cable 32.

[0083] The light receiver 27 is configured to decompose the reflected light by wavelength, and measure the intensity of the reflected light at each wavelength over a prescribed wavelength range. That is, the light receiver 27 decomposes the reflected light from the work W by wavelength, measures the intensity of the reflected light at each wavelength over a prescribed wavelength range, and thereby generates light intensity measurement data. The intensity of the reflected light at each wavelength can also be expressed as a relative value such as reflectance or relative reflectance. The light intensity measurement data is transmitted to the processing system 30.

[0084] The processing system 30 generates a film thickness map from the light intensity measurement data, as shown in FIG. 6, for example. Figure 3The spectrum of the reflected light is shown. In the following description, the spectrum of the reflected light from the workpiece W is referred to as a measurement spectrum. The measurement spectrum of the reflected light from the workpiece W contains information on the film thickness of the workpiece W. In other words, the measurement spectrum of the reflected light varies depending on the film thickness of the workpiece W. The processing system 30 is configured to determine the film thickness of the workpiece W based on the measurement spectrum of the reflected light. For example, the processing system 30 determines a reference spectrum that is closest in shape to the measurement spectrum of the reflected light from among a reference spectrum library, and determines a film thickness associated with the determined reference spectrum. In another example, the processing system 30 calculates a feature quantity of the measurement spectrum of the reflected light, determines a reference feature quantity that is closest to the feature quantity from among a reference feature quantity library, and determines a film thickness associated with the determined reference feature quantity. In another example, the processing system 30 performs Fourier transform on the measurement spectrum of the reflected light, and determines a film thickness from the obtained frequency spectrum.

[0085] Figure 4 is a diagram that shows an example of a plurality of film thickness measurement points on the workpiece W. As described above, in the polishing of the workpiece W, the optical sensor head 25 irradiates light to the surface of the workpiece W a plurality of times while moving across the surface of the workpiece W each time the polishing table 3 rotates one revolution. Therefore, as shown in Figure 4 , a plurality of film thickness measurement points M that are irradiated with light from the optical sensor head 25 are arranged in the radial direction on the surface of the workpiece W. Each time the polishing table 3 rotates one revolution, the optical sensor head 25 receives reflected light from the plurality of film thickness measurement points M, and the processing system 30 generates a plurality of measurement spectra of the reflected light from the plurality of film thickness measurement points M. Further, the processing system 30 determines a plurality of film thicknesses at the plurality of film thickness measurement points M from the plurality of measurement spectra of the reflected light.

[0086] The film thickness of the workpiece W varies depending on the measurement spectrum of the reflected light. Therefore, in order for the optical film thickness measurement device 20 to correctly measure the film thickness of the workpiece W, it is necessary to acquire a correct measurement spectrum that reflects the film thickness. However, due to failure or aging deterioration of optical elements such as the light source 22 and the light-irradiating optical fiber cable 31 and the light-receiving optical fiber cable 32, sometimes the measurement spectrum changes.

[0087] Therefore, as described below, the processing system 30 checks the measurement spectrum of the reflected light from the workpiece W. First, the processing system 30 polishes a reference workpiece that has the same surface configuration as the workpiece W using the polishing device shown in Figure 1 and Figure 2 . The reference workpiece is polished under the same polishing conditions as the workpiece W. The polishing conditions include the rotational speed of the polishing table 3, the rotational speed of the polishing head 1, the supply flow rate of the polishing liquid, the pressing pressure of the polishing head 1 against the workpiece W, and the like.

[0088] The processing system 30 determines a reference spectrum that is closest in shape to the measurement spectrum of the reflected light from the workpiece W from among the reference spectrum library, and determines a film thickness associated with the determined reference spectrum. In another example, the processing system 30 calculates a feature quantity of the measurement spectrum of the reflected light, determines a reference feature quantity that is closest to the feature quantity from among a reference feature quantity library, and determines a film thickness associated with the determined reference feature quantity. In another example, the processing system 30 performs Fourier transform on the measurement spectrum of the reflected light, and determines a film thickness from the obtained frequency spectrum. Figures 1 to 4The spectrum of the reflected light from the reference workpiece is generated by the method explained above. In the following explanation, the spectrum of the reflected light from the reference workpiece is referred to as a reference spectrum. As explained above with reference to FIG. 2, the reference spectrum is generated for each of the plurality of film thickness measurement points of the reference workpiece. Figure 4 As explained above with reference to FIG. 2, the reference spectrum is generated for each of the plurality of film thickness measurement points of the reference workpiece.

[0089] The processing system 30 calculates a feature quantity for each of the plurality of reference spectra. The feature quantity is an index indicating a feature of each reference spectrum. More specifically, as explained above with reference to FIG. 2, each reference spectrum indicates the intensity of the reflected light at each wavelength, and thus the feature quantity is an index indicating a feature of the intensity of the reflected light at each wavelength of the reference spectrum. Figure 3

[0090] In the present embodiment, the processing system 30 performs principal component analysis on the data set including the plurality of intensities of the reflected light at the plurality of wavelengths of each reference spectrum, and determines a feature quantity including at least the kth principal component (k is a natural number) obtained from the principal component analysis. One example of the calculation formula of the kth principal component is as follows:

[0091] kth principal component = w k1 X1 + w k2 X2 + w k2 X3 +...

[0092] Here, Xn (n is a natural number) indicates the intensity of the reflected light at the wavelength λn of the reference spectrum, and w kn is a weight coefficient with respect to the intensity Xn.

[0093] The processing system 30 calculates at least one feature quantity for each reference spectrum. For example, the processing system 30 calculates the kth principal component as the feature quantity for each reference spectrum. In another example, the processing system 30 can also calculate the kth principal component and the (k+1)th principal component as the feature quantity for each reference spectrum. In the present embodiment, the feature quantity is the kth principal component (k is a natural number) obtained by the principal component analysis of the data set, but in other embodiments, the feature quantity can also be a statistical value (for example, mean, standard deviation, variance, etc.) of the data set.

[0094] ​The optical sensor head 25 irradiates light to the plurality of film thickness measurement points on the reference workpiece and receives reflected light from the plurality of film thickness measurement points at each time interval within the polishing time of the reference workpiece. In one embodiment, the above-mentioned time interval is the time in which the polishing table 3 rotates L times (L is a natural number). That is, each time the polishing table 3 rotates L times, the optical sensor head 25 irradiates light to the plurality of film thickness measurement points on the reference workpiece and receives reflected light from the plurality of film thickness measurement points. Each time the polishing table 3 rotates L times, the processing system 30 generates a plurality of reference spectra of the reflected light from the plurality of film thickness measurement points on the reference workpiece and calculates a plurality of characteristic amounts from the plurality of reference spectra. Thus, the plurality of characteristic amounts are acquired each time the polishing table 3 rotates L times. In other embodiments, the above-mentioned time interval can also be defined by time. For example, the above-mentioned time interval can also be an interval defined in seconds as a unit of time (for example, 0.5 seconds, 1 second, or 1.5 seconds).

[0095] The processing system 30 associates the plurality of characteristic amounts with the plurality of times at which the plurality of reference spectra were generated, respectively. Each characteristic amount is calculated from the corresponding reference spectrum, and thus each characteristic amount can be associated with the time at which the corresponding reference spectrum was generated. The time at which the reference spectrum was generated is not particularly limited as long as it directly or indirectly indicates the time at which the reference spectrum was generated by the processing system 30. For example, the time at which the reference spectrum was generated can be indicated by the time at which the light used to generate the reference spectrum was emitted from the light source 22, or can be indicated by the time at which the spectrometer 27 received the reflected light used to generate the reference spectrum, or can be indicated by the time at which the processing system 30 received the light intensity measurement data used to generate the reference spectrum, or can be a time that is linked to the time at which the reference spectrum was generated by the processing system 30.

[0096] The processing system 30 performs data mapping by plotting the plurality of times at which the plurality of reference spectra were generated and the plurality of reference data points determined from the plurality of characteristic amounts corresponding to the plurality of times, respectively, on a coordinate system. Figure 5 is a diagram indicating one embodiment of data mapping. As shown in Figure 5 , the coordinate system has a vertical axis indicating the characteristic amount of the reference spectrum and a horizontal axis indicating the time.

[0097] The processing system 30 generates a plurality of reference spectra of the reflected light from the plurality of film thickness measurement points on the reference workpiece at each time interval within the polishing time of the reference workpiece and calculates a plurality of characteristic amounts of the plurality of reference spectra. In the embodiment described below, the above-mentioned time interval is the time in which the polishing table 3 rotates one time. Each time the polishing table 3 rotates one time, the processing system 30 plots the plurality of characteristic amounts of the plurality of reference spectra and the plurality of reference data points determined from the corresponding plurality of times on a coordinate system.

[0098] Figure 5 FDN This represents multiple reference data points acquired when the grinding table 3 rotates for the Nth time (N is a natural number), FD N+1 This represents multiple reference data points acquired when the grinding table 3 rotates for the (N+1)th revolution, FD N+2 This represents multiple reference data points acquired when the grinding table 3 rotates for the (N+2)th revolution. Figure 5 It can be seen that the number of reference data points increases with the grinding time of the reference workpiece. In the following description, the reference data point FD is sometimes referred to as... N FD N+1 FD N+2 …referred to as reference data point FD respectively.

[0099] The processing system 30 determines a threshold range for each time interval. The threshold range is the permissible dispersion range of multiple reference data points determined on a coordinate system for each time interval. All or most of the multiple reference data points FD acquired for each time interval fall within the corresponding threshold range. In one embodiment, a predetermined percentage (%) of the multiple reference data points FD fall within the corresponding threshold range. The predetermined percentage (%) is, for example, in the range of 90% to 100%.

[0100] exist Figure 5 In the example shown, the threshold range TR N FD from multiple reference data points N Determined, threshold range TR N+1 FD from multiple reference data points N+1 Determined, threshold range TR N+2 FD from multiple reference data points N+2 Okay. Similarly, processing system 30 determines multiple threshold ranges corresponding to multiple time intervals. In the following description, the threshold range TR is sometimes referred to as... N TR N+1 TR N+2 …referred to as threshold range TR respectively.

[0101] In one implementation, each threshold range TR is a range within a predetermined Mahalanobis distance from a reference point determined based on a plurality of reference data points FD within the corresponding time interval. The predetermined Mahalanobis distance is the distance between the plurality of reference data points FD that exist within the corresponding threshold range TR at the aforementioned predetermined percentage (%).

[0102] In another embodiment, each threshold range TR is a range within a prescribed Euclidean distance from a reference point determined from the plurality of reference data points FD within the corresponding time interval. The prescribed Euclidean distance is a distance within which the plurality of reference data points FD exist within the corresponding threshold range TR in the prescribed percentage (%). In yet another embodiment, each threshold range TR is a range determined based on a standard deviation or variance calculated from a normal distribution of the plurality of reference data points FD within the corresponding time interval.

[0103] As an example of the reference point, a center of gravity of the plurality of reference data points FD within the time interval corresponding to each threshold range TR can be cited. In one example, the center of gravity of the plurality of reference data points FD can also be calculated by multiplying the distances of the plurality of reference data points FD by weight coefficients that differ depending on the distances from the temporary center of gravity of the plurality of reference data points FD. The processing system 30 stores the plurality of threshold ranges TR corresponding to the plurality of time intervals in the storage device 30a.

[0104] In the above embodiment, the plurality of threshold ranges TR is determined from the reference data points obtained in the polishing of one reference workpiece, but in one embodiment, a plurality of reference workpieces can be polished in the same manner, and the plurality of threshold ranges TR corresponding to the plurality of time intervals can be determined from the reference data points obtained in the polishing of these reference workpieces.

[0105] The processing system 30 uses the threshold ranges TR to examine the measured spectrum of the reflected light generated in the polishing of the workpiece W. The polishing of the workpiece W and the generation of the measured spectrum of the reflected light from the workpiece W are performed in the same manner as the reference workpiece. That is, the workpiece W is polished while the polishing table 3 is rotated by the table motor 6 and the workpiece W is pressed against the polishing pad 2 on the polishing table 3 by the polishing head 1. The optical sensor head 25 irradiates light to the plurality of film thickness measurement points on the workpiece W at each time interval in the polishing of the workpiece W, and receives the reflected light from the plurality of film thickness measurement points. In one example, the time interval corresponds to the time for one rotation of the polishing table 3.

[0106] The processing system 30 generates a plurality of measurement spectra of the reflected light from the plurality of film thickness measurement points, and calculates a plurality of characteristic amounts of the plurality of measurement spectra. The characteristic amounts of the measurement spectra are calculated in the same manner as the characteristic amounts of the reference spectra. That is, the processing system 30 calculates the kth principal component (k is a natural number) by performing principal component analysis on a data set including a plurality of intensities of the reflected light at a plurality of wavelengths of each measurement spectrum, and determines a characteristic amount including at least the kth principal component. The characteristic amount of each measurement spectrum is an index of the same type as the characteristic amount of the reference spectra. That is, in a case where the characteristic amount of the reference spectra includes only the kth principal component, the characteristic amount of the measurement spectra also includes only the kth principal component. In a case where the characteristic amount of the reference spectra includes the kth principal component and the k+1th principal component, the characteristic amount of the measurement spectra also includes the kth principal component and the k+1th principal component.

[0107] The processing system 30 performs data mapping by plotting a plurality of times at which the plurality of measurement spectra are generated and a plurality of measurement data points determined from the plurality of characteristic amounts on a coordinate system. The data mapping is also performed in the same manner as the data mapping performed with respect to the reference spectra.

[0108] Figure 6 is a graph indicating a plurality of measurement data points MD N acquired when the polishing table 3 is rotated for the Nth time, Figure 7 is a graph indicating a plurality of measurement data points MD N+1 acquired when the polishing table 3 is rotated for the N+1th time, Figure 8 is a graph indicating a plurality of measurement data points MD N+2 acquired when the polishing table 3 is rotated for the N+2th time. In the following description, the measurement data points MD N , MD N+1 , MD N+2 … are sometimes simply referred to as measurement data points MD, respectively. As shown in Figures 6 to 8 , the processing system 30 plots a plurality of measurement data points MD on a coordinate system every time the polishing table 3 is rotated one time. The coordinate system has a vertical axis indicating a characteristic amount and a horizontal axis indicating a time.

[0109] The processing system 30 determines a correction index value indicating the number of the plurality of measurement data points MD present within the threshold value range TR determined on the coordinate system for each time interval. The correction index value is an index indicating the degree to which the plurality of measurement data points MD are present within the threshold value range TR.

[0110] As examples of the correction index value, the following examples can be given:

[0111] the number of the plurality of measurement data points present within the threshold value range;

[0112] The ratio of the number of multiple measurement data points existing within the threshold range to the total number of multiple measurement data points;

[0113] The distance between the centroid of multiple measured data points and the center of the threshold range.

[0114] In the embodiments described below, the correction index value is the ratio of the number of multiple measurement data points MD present within the threshold range TR to the total number of multiple measurement data points MD. Figure 6 In the example shown, the processing system 30 calculates the MD values ​​obtained from multiple measurement data points. N The threshold range TR corresponding to the time interval N Multiple measurement data points MD within N The number relative to multiple measurement data points MD N The proportion of the total number determines the correction index value for the calculated proportion.

[0115] In this embodiment, the correction index value is the proportion of multiple measurement data points MD that exist within the threshold range TR. For example, if the calculated correction index value is 100%, then all multiple measurement data points MD exist within the threshold range TR. If the calculated correction index value is 50%, then half of the multiple measurement data points MD exist within the threshold range TR. The processing system 30 determines whether the correction index value is within a predetermined correction management range. As described later, the correction management range is a numerical range used to determine whether to correct the measurement spectrum, use it directly for film thickness measurement without correcting the measurement spectrum, or generate an alarm signal. For example, the correction management range is 20% to 80%.

[0116] exist Figure 6 In the example shown, processing system 30 determines that the correction index value is greater than the upper limit of the correction management range, which is within the threshold range TR. N Multiple measurement data points MD within N The number relative to multiple measurement data points MD N The proportion of the total. The fact that the correction index value is larger than the upper limit of the correction management range means that multiple measurement data points MD N All or most of them exist within the threshold range TR N Therefore, when the calibration index value is larger than the upper limit of the calibration management range, the processing system 30 decides to use multiple measurement data points MD. N The corresponding multiple measurement spectra are normal. Based on these multiple measurement spectra, the multiple film thicknesses at multiple film thickness measurement points of workpiece W are determined (estimated).

[0117] Figure 6 The measurement data points MD shown NThis indicates that the corresponding measured spectrum is normal. Therefore, the measured data point MD N Capable of being used for threshold range TR N The update. Specifically, when the calibration index value is larger than the upper limit of the calibration management range, the processing system 30 updates multiple measurement data points MD. N Add to multiple reference data points FD N (Refer to Figure 5 To update multiple reference data points FD N Based on the updated multiple reference data points FD N Update threshold range TR N Because TR is within the threshold range N The reference data point FD used in the decision N With the increase in the number of values, the processing system 30 is able to determine a more accurate threshold range TR.

[0118] exist Figure 7 In the example shown, the processing system 30 determines that the correction index value is greater than the upper limit of the correction management range, which is within the threshold range TR. N+1 Multiple measurement data points MD within N+1 The number relative to multiple measurement data points MD N+1 The proportion of the total number. Therefore, the processing system 30 determines the ratio of multiple measured data points MD. N+1 The corresponding multiple measurement spectra are normal. Based on these multiple measurement spectra, the film thickness at multiple film thickness measurement points on workpiece W is determined (estimated). Because... Figure 7 The measurement data points MD shown N+1 This indicates that the corresponding measured spectrum is normal, therefore the measured data point MD N+1 Capable of being used for threshold range TR N+1 Update.

[0119] exist Figure 8 In the example shown, the processing system 30 determines that the correction index value is within the correction management range, which is within the threshold range TR. N+2 Multiple measurement data points MD within N+2 The number relative to multiple measurement data points MD N+2 The proportion of the total. When the correction index value is within the correction management range, such as Figure 9 As shown, the processing system 30 enables multiple measurement data points MD N+2 Move along the coordinate system until multiple measured data points MD are reached. N+2 Converging to the threshold range TR N+2 Up to this point, based on multiple measurement data points MD N+2 Correction is performed on multiple identified characteristic quantities. Multiple measurement data points MD N+2Examples of movement include parallel movement, rotation, and shrinking.

[0120] In one implementation, the processing system 30 will move multiple measurement data points MD N+2 The corrected feature quantities are then converted into multiple corrected measurement spectra. In this embodiment, each measurement data point MD N+2 The eigenvalues ​​are obtained through principal component analysis, specifically the k-th principal component, thus allowing the k-th principal component to be restored to the original dataset form. Specifically, the processing system 30 converts the corrected eigenvalues ​​into multiple intensities of reflected light at multiple wavelengths, and generates a corrected measurement spectrum based on the converted intensities and corresponding wavelengths. The corrected measurement spectrum is generated within the threshold range TR. N+2 The system generates measurement data points that exist within the system, so the processing system 30 can determine the accurate film thickness based on the corrected measurement spectrum.

[0121] In this embodiment, the feature quantity is the k-th principal component (k is a natural number) obtained through principal component analysis of the dataset. However, in other embodiments, the feature quantity can also be a statistical value of the dataset (e.g., mean, standard deviation, variance, etc.). When the feature quantity is the average of multiple intensities contained in the dataset, the multiple intensities of reflected light at multiple wavelengths are respectively subtracted by the intensity equivalent to the shift of the measured data point towards the threshold range TR, thereby correcting the multiple intensities. The corrected measurement spectrum is generated based on the corrected multiple intensities and the corresponding multiple wavelengths.

[0122] In one embodiment, the processing system 30 may also calculate the film thickness of the workpiece W based on the corrected characteristic quantities instead of generating the corrected measurement spectrum. For example, the processing system 30 determines the reference characteristic quantity closest to the corrected characteristic quantity from a reference characteristic quantity library, and determines the film thickness associated with the determined reference characteristic quantity. The reference characteristic quantity library contains multiple reference characteristic quantities and multiple corresponding film thicknesses.

[0123] exist Figure 10 In the example shown, the processing system 30 determines that the correction index value is less than the lower limit of the correction management range, which is within the threshold range TR. N+3 Multiple measurement data points MD within N+3 The number relative to multiple measurement data points MD N+3 The proportion of the total. The fact that the correction index value is less than the lower limit of the correction management range means that multiple measurement data points MD N+3 All or most of them exist within the threshold range TR N+3 Therefore, when the correction index value is less than the lower limit of the correction management range, the processing system 30 generates a representation of multiple measurement data points MD. N+3 The corresponding alarm signal for abnormal measurement spectrum.

[0124] In one embodiment, the processing system 30 can also transmit the alarm signal to the action control section 9, which gives an instruction to the polishing device to end polishing of the workpiece W when receiving the alarm signal. The action control section 9 can also give an instruction to the polishing device to perform re-polishing of the workpiece W. In other embodiments, the processing system 30 can also transmit the alarm signal to a host computer that manages the polishing device. By transmitting the alarm signal to the host computer, the manager can be prompted to perform inspection or repair of the polishing device.

[0125] Figure 11 is a flowchart of the embodiment described above with reference to Figures 1 to 10

[0126] In step S101, the reference workpiece is polished while the polishing table 3 is rotated and the reference workpiece is pressed against the polishing pad 2 on the polishing table 3. At each time interval within the polishing time of the reference workpiece, the optical sensor head 25 irradiates light to a plurality of film thickness measurement points on the reference workpiece, and the processing system 30 generates a plurality of reference spectra of the reflected light from the plurality of film thickness measurement points.

[0127] The plurality of film thickness measurement points of the reference workpiece are distributed over the entire surface of the reference workpiece. The device region and the dicing groove are mixedly present on the surface of the reference workpiece. The reflected light from the dicing groove is significantly different from the reflected light from the device region, and can become noise for film thickness measurement. Therefore, in one embodiment, the processing system 30 can also perform filtering processing on the plurality of reference spectra generated in the polishing of the reference workpiece to remove the reference spectra that can become noise.

[0128] In step S102, the processing system 30 calculates a plurality of characteristic amounts of the plurality of reference spectra. At least one characteristic amount is calculated for each reference spectrum. As examples of the characteristic amount, the k-th principal component (k is a natural number) obtained by principal component analysis, and a statistical value such as an average can be given.

[0129] In step S103, the processing system 30 performs data mapping by plotting a plurality of times at which the plurality of reference spectra are generated and a plurality of reference data points FD determined from the corresponding plurality of characteristic amounts on a coordinate system (refer to Figure 5 ).

[0130] In step S104, the processing system 30 determines a threshold range TR for each time interval within the polishing time of the reference workpiece on the basis of the plurality of reference data points FD on the coordinate system (refer to Figure 5 ​). In one embodiment, the time interval is the time for the polishing table 3 to rotate one revolution. A plurality of threshold ranges TR corresponding to the plurality of time intervals are determined. In one embodiment, each threshold range TR is a range within a prescribed Mahalanobis distance from a reference point determined in advance based on the plurality of reference data points FD within the corresponding time interval.

[0131] In step S105, the workpiece W is polished while the polishing table 3 is rotated and the workpiece W is pressed against the polishing pad 2 on the polishing table 3. At each time interval in the polishing of the workpiece W, the optical sensor head 25 irradiates light toward a plurality of film thickness measurement points on the workpiece W, and the processing system 30 generates a plurality of measurement spectra of the reflected light from the plurality of film thickness measurement points. In one embodiment, the processing system 30 can also perform filtering processing on the plurality of measurement spectra generated in the polishing of the workpiece W to remove measurement spectra that can be noise.

[0132] In step S106, the processing system 30 calculates a plurality of feature amounts of the plurality of measurement spectra. The feature amounts of the measurement spectra are calculated by the same method as the feature amounts of the reference spectra calculated in step S102 described above.

[0133] In step S107, the processing system 30 performs data mapping by plotting the plurality of times at which the plurality of measurement spectra were generated and the plurality of measurement data points MD determined from the plurality of feature amounts of the plurality of measurement spectra on the coordinate system (refer to FIG. 6). Figure 6 、 Figure 7 、 Figure 8 ).

[0134] In step S108, the processing system 30 determines a correction index value indicating the number of the plurality of measurement data points present within each threshold range TR determined on the coordinate system for each time interval, and determines whether the correction index value is within a correction management range. In one example, the correction index value is the proportion of the number of the plurality of measurement data points present within each threshold range TR to the total number of the plurality of measurement data points. In another example, the correction index value is the number of the plurality of measurement data points present within each threshold range TR itself.

[0135] In step S109, when the correction index value is within the correction management range, the processing system 30 moves the plurality of measurement data points MD on the coordinate system until the plurality of measurement data points MD converge within the corresponding threshold range TR, thereby correcting the plurality of feature amounts corresponding to the plurality of measurement data points MD, respectively (refer to FIG. 6). Figure 9). In one embodiment, the processing system 30 generates the plurality of corrected measurement spectra by converting the plurality of corrected characteristic quantities into the plurality of measurement spectra. The processing system 30 determines (estimates) the plurality of film thicknesses at the plurality of film thickness measurement points of the workpiece W from the plurality of corrected measurement spectra. In other embodiments, the processing system 30 does not generate the corrected measurement spectra, but determines (estimates) the plurality of film thicknesses at the plurality of film thickness measurement points of the workpiece W from the plurality of corrected characteristic quantities.

[0136] In step S110, when the correction index value is greater than the upper limit of the correction management range, the processing system 30 does not correct the plurality of characteristic quantities corresponding to the plurality of measurement data points MD, but determines (estimates) the plurality of film thicknesses at the plurality of film thickness measurement points of the workpiece W from the measurement spectra.

[0137] In step S111, when the correction index value is less than the lower limit of the correction management range, the processing system 30 generates an alarm signal. The processing system 30 can also transmit the alarm signal to the action control section 9 to stop the polishing of the workpiece W.

[0138] The spectrum of the reflected light shows different tendencies depending on the surface configuration of the workpiece W as the polishing time changes. In particular, the spectrum of the reflected light from the workpiece W is not only affected by the uppermost film of the workpiece W, but also by the lower layer present thereunder. As a result, the measurement data points determined from the characteristic quantities and the time can change depending on the surface configuration of the workpiece W as the polishing time progresses. In order to correctly examine the spectrum of the reflected light from the workpiece W, it is necessary to select an appropriate threshold range based on the difference in the surface configuration of the workpiece W.

[0139] Therefore, with reference to Figure 12 In the embodiments described below, the processing system 30 has a first group and a second group of a plurality of threshold ranges determined from the reference data points obtained in the polishing of a plurality of reference workpieces having different surface configurations. The details of the present embodiments not specifically described are the same as those of the previously described embodiments, and therefore the repeated description thereof is omitted.

[0140] The first group of the plurality of threshold ranges is made from the reference data points obtained in the polishing of a first reference workpiece, and the second group of the plurality of threshold ranges is made from the reference data points obtained in the polishing of a second reference workpiece. The first reference workpiece and the second reference workpiece are polished under the same polishing conditions as the workpiece W. The first group and the second group of the plurality of threshold ranges are made in the same manner as the reference Figure 5 The first group and the second group of the plurality of threshold ranges are made in the same manner as the reference

[0141] The first reference workpiece and the second reference workpiece have different surface configurations. As a result, as Figure 12As shown, the tendency of multiple threshold ranges TR1 in the first group to change with grinding time differs from the tendency of multiple threshold ranges TR2 in the second group to change with grinding time. Figure 12 In the example shown, the multiple threshold ranges TR1 in the first group show a decreasing trend with grinding time, while the multiple threshold ranges TR2 in the second group show an increasing trend with grinding time. The first group of multiple threshold ranges TR1 and the second group of multiple threshold ranges TR2 are stored in the storage device 30a of the processing system 30.

[0142] Figure 13 This is a graph illustrating the time progression of measurement data points MD plotted on a coordinate system during the grinding of workpiece W. The processing system 30 selects data points MD based on their tendency to shift over time with the grinding period of workpiece W. Figure 12 Either the first group of multiple threshold ranges TR1 or the second group of multiple threshold ranges TR2 shown. Figure 13 In the example shown, the measured data point MD shows an increasing trend with grinding time. Therefore, the processing system 30 selects a second group of multiple threshold ranges TR2. After selecting the second group, the processing system 30 determines a correction index value in the same manner as in the above-described embodiment, wherein the correction index value represents the number of multiple measured data points MD existing within the threshold range TR2 determined on the coordinate system for each time interval, and determines whether the correction index value is within the correction management range.

[0143] In one embodiment, the processing system 30 may also have multiple threshold ranges from three or more groups based on reference data points obtained during the grinding of three or more reference workpieces with different surface textures. The processing system 30 may also select one group from the three or more groups based on the tendency of multiple measured data points as the grinding time of the workpiece W progresses.

[0144] Figure 14 It is a reference Figure 12 and Figure 13 A flowchart illustrating the described implementation method.

[0145] In step S201, a first reference workpiece is polished while the polishing table 3 is rotated and the first reference workpiece is pressed against the polishing pad 2 on the polishing table 3. At each time interval during the polishing of the first reference workpiece, the optical sensor head 25 irradiates light to a plurality of film thickness measurement points on the first reference workpiece, and the processing system 30 generates a plurality of reference spectra of the reflected light from the plurality of film thickness measurement points. Likewise, a second reference workpiece is polished while the polishing table 3 is rotated and the second reference workpiece is pressed against the polishing pad 2 on the polishing table 3. At each time interval during the polishing of the second reference workpiece, the optical sensor head 25 irradiates light to a plurality of film thickness measurement points on the second reference workpiece, and the processing system 30 generates a plurality of reference spectra of the reflected light from the plurality of film thickness measurement points.

[0146] In step S202, the processing system 30 calculates a plurality of characteristic amounts of the plurality of reference spectra obtained from the polishing of the first reference workpiece, and a plurality of characteristic amounts of the plurality of reference spectra obtained from the polishing of the second reference workpiece.

[0147] In step S203, the processing system 30 performs data mapping by plotting a plurality of times at which the plurality of reference spectra obtained from the polishing of the first reference workpiece are generated and a plurality of reference data points FD determined from the corresponding plurality of characteristic amounts on a coordinate system (refer to FIG. 6). Likewise, data mapping is performed by plotting a plurality of times at which the plurality of reference spectra obtained from the polishing of the second reference workpiece are generated and a plurality of reference data points FD determined from the corresponding plurality of characteristic amounts on a coordinate system (refer to FIG. 7). Figure 12 ). Likewise, data mapping is performed by plotting a plurality of times at which the plurality of reference spectra obtained from the polishing of the second reference workpiece are generated and a plurality of reference data points FD determined from the corresponding plurality of characteristic amounts on a coordinate system (refer to FIG. 7). Figure 12

[0148] In step S204, the processing system 30 determines a first group of a plurality of threshold ranges TR1 determined at each time interval within the polishing time of the first reference workpiece, on the basis of the plurality of reference data points FD on the coordinate system (refer to FIG. 6). Likewise, the processing system 30 determines a second group of a plurality of threshold ranges TR2 determined at each time interval within the polishing time of the second reference workpiece, on the basis of the plurality of reference data points FD on the coordinate system (refer to FIG. 7). Figure 12 ). Likewise, the processing system 30 determines a second group of a plurality of threshold ranges TR2 determined at each time interval within the polishing time of the second reference workpiece, on the basis of the plurality of reference data points FD on the coordinate system (refer to FIG. 7). Figure 12

[0149] In step S205, a workpiece W is polished while the polishing table 3 is rotated and the workpiece W is pressed against the polishing pad 2 on the polishing table 3. At each time interval during the polishing of the workpiece W, the optical sensor head 25 irradiates light to a plurality of film thickness measurement points on the workpiece W, and the processing system 30 generates a plurality of measurement spectra of the reflected light from the plurality of film thickness measurement points.

[0150] In step S206, the processing system 30 calculates a plurality of characteristic amounts of the plurality of measurement spectra.

[0151] ​​In step S207, the processing system 30 performs data mapping by plotting the plurality of time at which the plurality of measured spectra were generated and the plurality of measured data points MD determined from the plurality of features of the plurality of measured spectra on a coordinate system (refer to Figure 13 ).

[0152] In step S208, the processing system 30 selects either the first group of the plurality of threshold ranges TR1 or the second group of the plurality of threshold ranges TR2 according to the progress of the plurality of measured data points MD with the polishing time of the workpiece W.

[0153] In step S209, the processing system 30 determines a correction index value that indicates the number of the plurality of measured data points MD present within the selected threshold range TR1 of the first group or the threshold range TR2 of the second group, and determines whether the correction index value is within a predetermined correction management range.

[0154] In step S210, when the correction index value is within the correction management range, the processing system 30 moves the plurality of measured data points MD on the coordinate system until the plurality of measured data points MD converge within the corresponding threshold range TR1 or TR2, thereby correcting the plurality of features corresponding to the plurality of measured data points MD (refer to Figure 9 ). In one embodiment, the processing system 30 generates a plurality of corrected measured spectra by converting the plurality of corrected features into the plurality of measured spectra. The processing system 30 determines (estimates) the plurality of film thicknesses at the plurality of film thickness measurement points of the workpiece W from the plurality of corrected measured spectra. In other embodiments, the processing system 30 does not generate corrected measured spectra, but determines (estimates) the plurality of film thicknesses at the plurality of film thickness measurement points of the workpiece W from the plurality of corrected features.

[0155] In step S211, when the correction index value is greater than the upper limit of the correction management range, the processing system 30 does not correct the plurality of features corresponding to the plurality of measured data points MD, but determines (estimates) the plurality of film thicknesses at the plurality of film thickness measurement points of the workpiece W from the measured spectra.

[0156] In step S212, when the correction index value is less than the lower limit of the correction management range, the processing system 30 generates an alarm signal. The processing system 30 can also send the alarm signal to the action control section 9 to stop the polishing of the workpiece W.

[0157] In the embodiment described with reference to Figures 12 to 14 , a plurality of groups of the plurality of threshold ranges prepared in advance are used. In one embodiment, the processing system 30 can also make a new group of the plurality of threshold values from the measured spectra obtained in the polishing of the workpiece W. For example, the processing system 30 at an initial stage of the polishing of the workpiece W, the processing system 30 has onlyFigure 12 The first group of multiple threshold ranges TR1 shown is constructed based on the measured spectra obtained during the grinding of workpiece W. Figure 12 The second group of multiple threshold ranges TR2 shown.

[0158] At multiple measurement data points of MD as the workpiece grinding time increases, such as... Figure 13 As shown, in cases where an upward trend is observed, multiple measured data points are at MD. Figure 12 The first group of thresholds falls outside the range TR1. Therefore, the processing system 30 generates an alarm signal, instructing the grinding device to stop grinding the workpiece. However, Figure 13 The upward trend shown is caused by a different type of workpiece than the one that should have been ground, not by a malfunction of the grinding equipment.

[0159] Therefore, in this embodiment, during the grinding of workpiece W, when the correction index value is less than the lower limit of the correction management range in multiple consecutive time intervals, the processing system 30 creates multiple new threshold ranges on the coordinate system corresponding to the aforementioned multiple consecutive time intervals based on multiple measurement data points obtained in these multiple consecutive time intervals.

[0160] For example, such as Figure 15 As shown, multiple measurement data points MD were obtained during consecutive time intervals in the grinding of workpiece W. N MD N+1 MD N+2 MD N+3 Deviation from the preset threshold range TR1 N TR1 N+1 TR1 N+2 TR1 N+3 In this case, the processing system 30 calculates data based on multiple measurement data points MD. N MD N+1 MD N+2 MD N+3 …Create new threshold ranges TR2 respectively N TR2 N+1 TR2 N+2 TR2 N+3 …In one embodiment, the aforementioned consecutive time intervals represent the initial grinding time of workpiece W. When grinding the next workpiece, a newly created threshold range TR2 is used. N TR2 N+1 TR2 N+2 TR2 N+3 … The processing system 30 is able to accurately check the measured spectrum using the newly created threshold range.

[0161] Next, refer to Figures 16 to 18Another other embodiment of making the threshold range will be described. In this example, the first reference workpiece and the second reference workpiece having the same surface configuration are used to make the first threshold range and the second threshold range of each time interval in the polishing of the first reference workpiece and the second reference workpiece.

[0162] As Figure 16 shown, the first reference workpiece is polished by using the above-mentioned polishing apparatus, and a plurality of first threshold ranges TR1 (TR1 N , TR1 N+1 , TR1 N+2 …) corresponding to a plurality of time intervals in the polishing of the first reference workpiece are made from a plurality of reference data points FD (FD N , FD N+1 , FD N+2 …) obtained in the polishing of the first reference workpiece. Likewise, the second reference workpiece is polished by using the above-mentioned polishing apparatus, and a plurality of second threshold ranges TR2 (TR2 N , TR2 N+1 , TR2 N+2 …) corresponding to a plurality of time intervals in the polishing of the second reference workpiece are made from a plurality of reference data points FD (FD N , FD N+1 , FD N+2 …) obtained in the polishing of the second reference workpiece.

[0163] Figure 17 is a plan view showing an example of the positional relationship of the optical sensor head 25 and the polishing head 1. As Figure 17 shown, in the polishing of the first reference workpiece W1, the optical sensor head 25 irradiates light to a plurality of film thickness measurement points on the surface of the first reference workpiece W1 while traversing the surface of the first reference workpiece W1 with the rotation of the polishing table 3. Likewise, in the polishing of the second reference workpiece W2, the optical sensor head 25 irradiates light to a plurality of film thickness measurement points on the surface of the second reference workpiece W2 while traversing the surface of the second reference workpiece W2 with the rotation of the polishing table 3.

[0164] The relative position of the optical sensor head 25 with respect to the first reference workpiece W1 at the polishing start time point of the first reference workpiece W1 is sometimes different from the relative position of the optical sensor head 25 with respect to the second reference workpiece W2 at the polishing start time point of the second reference workpiece W2. That is, the time when the optical sensor head 25 initially traverses the surface of the first reference workpiece W1 is sometimes different from the time when the optical sensor head 25 initially traverses the surface of the second reference workpiece W2. For this reason, as Figure 16 shown, the positions of the plurality of threshold ranges TR1 on the time axis are sometimes different from the positions of the plurality of threshold ranges TR2 on the time axis.

[0165] Therefore, the processing system 30 will... Figure 16 The multiple first threshold ranges TR1 (TR1) shown N TR1 N+1 TR1 N+2 …) and multiple second threshold ranges TR2 (TR2) N TR2 N+1 TR2 N+2 …) combine them to create Figure 18 As shown, multiple threshold ranges (TR1) correspond to different time intervals. N TR2 N TR1 N+1 TR2 N+1 TR1 N+2 TR2 N+2 …). More specifically, by using multiple first threshold ranges TR1 (TR1 N TR1 N+1 TR1 N+2 …) and multiple second threshold ranges TR2 (TR2) N TR2 N+1 TR2 N+2 …) Arranged along the time axis, thereby combining the first threshold range TR1 and multiple second threshold ranges TR2.

[0166] The resulting multiple threshold ranges reflect the differences in the initial position of the optical sensor head 25, thus enabling the processing system 30 to use... Figure 18 The threshold range shown is used to correctly check the measured spectrum.

[0167] In this embodiment, two reference artifacts are used to create multiple threshold ranges corresponding to different time intervals, but three or more reference artifacts can also be used to create multiple threshold ranges corresponding to different time intervals.

[0168] Figure 19 This is a top view showing one embodiment of a polishing apparatus equipped with a first optical sensor head 25A and a second optical sensor head 25B. The first optical sensor head 25A and the second optical sensor head 25B are respectively connected to... Figure 2 The light source 22 and beam splitter 27 are optically connected. The first optical sensor head 25A and the second optical sensor head 25B are arranged at different positions circumferentially on the grinding table 3. The first optical sensor head 25A and the second optical sensor head 25B are located at the same distance from the center of the grinding table 3. Other structures of the grinding apparatus are similar to those shown in the reference diagram. Figure 1 and Figure 2 The implementation methods described are the same, so repeated descriptions are omitted.

[0169] likeFigure 19 As shown, during the grinding of workpiece W, the first optical sensor head 25A and the second optical sensor head 25B rotate together with the grinding table 3 while moving alternately (i.e. at different times) across the surface of workpiece W.

[0170] The threshold range is established based on reference data points obtained during the grinding of the reference workpiece Wr. During the grinding of the reference workpiece Wr, the first optical sensor head 25A and the second optical sensor head 25B rotate together with the grinding table 3 and move alternately (i.e. at different times) across the surface of the reference workpiece Wr.

[0171] Figure 20 This refers to multiple reference data points FD (FD) obtained during the grinding of the reference workpiece Wr by light irradiation from the first optical sensor head 25A. N FD N+1 FD N+2 ...) multiple first threshold ranges TR1 (TR1) were created N TR1 N+1 TR1 N+2 …) and multiple reference data points FD (FD) obtained by light irradiation from the second optical sensor head 25B during the grinding of the reference workpiece Wr. N FD N+1 FD N+2 ...) multiple second threshold ranges TR2 (TR2) were created N TR2 N+1 TR2 N+2 A diagram of an example of …).

[0172] like Figure 20 As shown, depending on the positions of the first optical sensor head 25A and the second optical sensor head 25B, the positions of multiple threshold ranges TR1 on the time axis differ from the positions of multiple threshold ranges TR2 on the time axis. Furthermore, due to individual differences between the first optical sensor head 25A and the second optical sensor head 25B, the positions of multiple threshold ranges TR1 on the feature axis differ from the positions of multiple threshold ranges TR2 on the feature axis.

[0173] Therefore, the processing system 30 will... Figure 20 The multiple first threshold ranges TR1 (TR1) shown N TR1 N+1 TR1 N+2 …) and multiple second threshold ranges TR2 (TR2) N TR2 N+1 TR2 N+2 …) are combined to create multiple threshold ranges (TR1) corresponding to different time intervals. NTR2 N TR1 N+1 TR2 N+1 TR1 N+2 TR2 N+2 ). More specifically, as shown in Figure 21 , the processing system 30 arranges the plurality of first threshold ranges TR1 and the plurality of second threshold ranges TR2 along the time axis, and further, as shown in Figure 22 , the processing system 30 arranges the plurality of first threshold ranges TR1 and the plurality of second threshold ranges TR2 in alignment by moving at least one of the plurality of first threshold ranges TR1 and the plurality of second threshold ranges TR2 along the characteristic amount axis.

[0174] The plurality of threshold ranges thus obtained reflect differences and individual differences in the configurations of the first optical sensor head 25A and the second optical sensor head 25B, and therefore the processing system 30 is able to correctly examine the measurement spectrum using the threshold ranges shown in Figure 22

[0175] In the present embodiment, the two first optical sensor head 25A and the second optical sensor head 25B are used to make the plurality of threshold ranges corresponding to different time intervals, but three or more optical sensor heads can be used to make the plurality of threshold ranges corresponding to different time intervals.

[0176] The processing system 30 acts in accordance with commands contained in a program stored in the storage device 30a. The program for causing the processing system 30 to execute the steps performed according to the embodiment described with reference to Figures 1 to 22 is recorded in a computer-readable recording medium as a non-transitory tangible object, and is supplied to the processing system 30 via the recording medium. Alternatively, the program can be input to the processing system 30 via a communication network such as the Internet or a local area network.

[0177] The above-described embodiments are described with the aim of enabling those skilled in the art to implement the application, and various changes in form that the application is amenable to are naturally possible without departing from the spirit of the application. Therefore, the technical scope of the application is not limited to the embodiments described above, but encompasses various modifications made on the basis of the above descriptions. The technical ideas of the application can also be applied to other embodiments. Therefore, the technical scope of the application is not limited to the embodiments described above, but encompasses the technical scope of the present application defined by the matters recited in the scope of claims.​

Claims

1. A method for examining the spectrum of reflected light from a workpiece, characterized in that, While the grinding table is rotating, the workpiece is pressed against the grinding pad on the grinding table to grind the workpiece. During each time interval of the grinding process on the workpiece, light is irradiated from the optical sensor head onto multiple film thickness measurement points on the workpiece. Multiple measurement spectra of reflected light from the multiple film thickness measurement points are generated. Calculate multiple characteristic quantities of the multiple measured spectra. Multiple times from the generated multiple measurement spectra, and multiple measurement data points determined based on the multiple characteristic quantities of the multiple measurement spectra, are plotted on a coordinate system, thereby performing data mapping. A correction index value is determined, which represents the number of the plurality of measured data points existing within a threshold range defined on the coordinate system for each time interval. Determine whether the correction index value is within the correction management range.

2. The method for examining the spectrum of reflected light from a workpiece according to claim 1, characterized in that, When the correction index value is within the correction management range, the plurality of measurement data points are moved on the coordinate system until the plurality of measurement data points converge within the threshold range, thereby correcting the plurality of feature quantities.

3. The method for examining the spectrum of reflected light from a workpiece according to claim 1, characterized in that, An alarm signal is generated when the value of the correction index is less than the lower limit of the correction management range.

4. The method for examining the spectrum of reflected light from a workpiece according to claim 1, characterized in that, It also includes the following steps: While rotating the grinding table, a reference workpiece is pressed against the grinding pad to grind the reference workpiece. During each time interval of the grinding of the reference workpiece, light is irradiated from the optical sensor head onto multiple film thickness measurement points on the reference workpiece. Multiple reference spectra of reflected light from the multiple film thickness measurement points are generated. Calculate multiple characteristic quantities of the multiple reference spectra. Multiple times from which the multiple reference spectra were generated, and multiple reference data points determined based on the multiple feature quantities of the multiple reference spectra, are plotted on the coordinate system, thereby performing data mapping. The threshold range for each time interval in the grinding of the reference workpiece is generated based on the plurality of reference data points on the coordinate system.

5. The method for examining the spectrum of reflected light from a workpiece according to claim 4, characterized in that, The threshold range is a range within a specified Mahalanobis distance from the reference point of the plurality of reference data points.

6. The method for examining the spectrum of reflected light from a workpiece according to claim 4, characterized in that, When the correction index value is greater than the upper limit of the correction management range, the plurality of reference data points are updated by adding the plurality of measurement data points to the plurality of reference data points, and the threshold range is updated based on the updated plurality of reference data points.

7. The method for examining the spectrum of reflected light from a workpiece according to claim 4, characterized in that, The method further includes the step of selecting, based on the progression of the plurality of measurement data points over the grinding time of the workpiece, any one of a first group of multiple threshold ranges and a second group of multiple threshold ranges. The threshold range is one of multiple threshold ranges from one of the selected groups in the first and second groups. The first set of the multiple threshold ranges is predetermined based on multiple reference data points obtained during the grinding of the first reference workpiece. The second set of the plurality of threshold ranges is predetermined based on a plurality of reference data points obtained during grinding of a second reference workpiece having a surface structure different from that of the first reference workpiece.

8. The method for examining the spectrum of reflected light from a workpiece according to claim 4, characterized in that, The reference workpiece is a first reference workpiece and a second reference workpiece. The method further includes the following step: creating multiple threshold ranges corresponding to different time intervals by combining a first threshold range created based on multiple reference data points obtained in the grinding of the first reference workpiece with a second threshold range created based on multiple reference data points obtained in the grinding of the second reference workpiece.

9. The method for examining the spectrum of reflected light from a workpiece according to claim 4, characterized in that, The optical sensor head is a first optical sensor head and a second optical sensor head disposed at different positions within the grinding table. The method further includes the following steps: creating multiple threshold ranges corresponding to different time intervals by combining a first threshold range and a second threshold range, wherein the first threshold range is created based on multiple reference data points obtained by light irradiation from the first optical sensor head during the grinding of the reference workpiece, and the second threshold range is created based on multiple reference data points obtained by light irradiation from the second optical sensor head during the grinding of the reference workpiece.

10. The method for examining the spectrum of reflected light from a workpiece according to claim 1, characterized in that, Each of the plurality of features contains at least a k-th principal component, which is obtained by performing principal component analysis on a dataset containing multiple intensities of the reflected light at multiple wavelengths of each measured spectrum, where k is a natural number.

11. The method for examining the spectrum of reflected light from a workpiece according to claim 1, characterized in that, In the grinding of the workpiece, during multiple consecutive time intervals, when the correction index value is smaller than the lower limit of the correction management range, multiple new threshold ranges corresponding to the multiple consecutive time intervals are created based on multiple measurement data points obtained in the multiple consecutive time intervals.

12. A grinding apparatus, a grinding apparatus for workpieces, characterized in that, have: Grinding table; A motor that rotates the grinding table; A grinding head that presses the workpiece onto a grinding pad on the grinding table to grind the workpiece; An optical sensor head that illuminates multiple film thickness measurement points on the workpiece at each time interval during the grinding of the workpiece; as well as The processing system generates multiple measurement spectra of reflected light from the multiple film thickness measurement points. The processing system is configured such that, Calculate multiple characteristic quantities of the multiple measured spectra. Multiple times from the generated multiple measurement spectra, and multiple measurement data points determined based on the multiple characteristic quantities of the multiple measurement spectra, are plotted on a coordinate system, thereby performing data mapping. A correction index value is determined, which represents the number of the plurality of measured data points existing within a threshold range defined on the coordinate system for each time interval. Determine whether the correction index value is within the correction management range.

13. The grinding apparatus according to claim 12, characterized in that, The processing system is configured to, when the correction index value is within the correction management range, move the plurality of measurement data points on the coordinate system until the plurality of measurement data points converge within the threshold range, thereby correcting the plurality of feature quantities.

14. The grinding apparatus according to claim 12, characterized in that, The processing system is configured to generate an alarm signal when the correction index value is less than the lower limit of the correction management range.

15. The grinding apparatus according to claim 12, characterized in that, The processing system is configured to select, based on the progression of the plurality of measured data points over the grinding time of the workpiece, either a first group or a second group of multiple threshold ranges. The threshold range is one of multiple threshold ranges from one of the selected groups in the first and second groups. The first set of the plurality of threshold ranges is predetermined based on reference data points obtained during the grinding of a first reference workpiece. The second set of the plurality of threshold ranges is predetermined based on reference data points obtained during grinding of a second reference workpiece having a surface structure different from that of the first reference workpiece.

16. The grinding apparatus according to claim 12, characterized in that, The processing system is configured such that, Principal component analysis was performed on the dataset containing multiple intensities of the reflected light at multiple wavelengths for each measured spectrum. The determination includes at least the feature quantity of the k-th principal component obtained from the principal component analysis, where k is a natural number.

17. The grinding apparatus according to claim 12, characterized in that, The processing system is configured to, during a series of consecutive time intervals in the grinding of the workpiece, when the correction index value is less than the lower limit of the correction management range, generate multiple new threshold ranges corresponding to the series of consecutive time intervals based on multiple measurement data points acquired during the series of consecutive time intervals.

18. The grinding apparatus according to claim 12, characterized in that, The processing system is configured to determine the threshold range, which is a range within a specified Mahalanobis distance, based on a reference point of multiple reference data points obtained from grinding a reference workpiece.

19. The grinding apparatus according to claim 18, characterized in that, The processing system is configured to update the plurality of reference data points by adding the plurality of measurement data points to the plurality of reference data points when the correction index value is greater than the upper limit of the correction management range, and to update the threshold range based on the updated plurality of reference data points.

Images