Grindstone evaluation method and grindstone evaluation device

Abstract

To properly evaluate a grinding wheel in a non-destructive manner.SOLUTION: Microscopic Raman spectroscopy is performed on the surface (working surface) of the grindstone 100. The scattered light S1 emitted from the sample (sample stage 90 side) passes through the objective lens 14, is reflected in the vertical direction by the beam splitter 13, passes through the pin hole 15, and is incident on the spectroscope 16 provided with a grating, where the spectrum of the scattered light S1 is measured. The analysis unit 20, which is a computer, controls the sample stage 90 and acquires a Raman spectrum (described later) at each point on the surface (working surface) of the grindstone 100 from the spectrometer 16. Thereafter, a feature amount as described later is extracted from the Raman spectrum, and the surface of the grindstone 100 is analyzed.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a method and an apparatus for evaluating a grinding wheel used for grinding silicon wafers and the like. [Background technology]

[0002] Silicon wafers, which are used as substrates for manufacturing semiconductor devices, are manufactured by cutting (slicing), grinding (grinding), or polishing (lapping, polishing) silicon ingots to produce wafers with high flatness. For this reason, grinding processing technology is important, and there is a need for evaluation techniques to determine whether the surface processing has been performed satisfactorily. As for grinding stones used for grinding semiconductor materials, those with a structure in which diamond abrasive grains are dispersed in a binder are particularly preferred.

[0003] Patent Document 1 describes a technique for using Raman spectroscopy to evaluate the surface alteration layer formed on the silicon surface during lapping. This technique can be used to evaluate the surface alteration layer and to appropriately remove it. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2023-72744 Summary of the Invention [Problem to be solved by the invention]

[0005] The technology described in Patent Document 1 allows for non-destructive and appropriate evaluation of the ground silicon wafer surface. Meanwhile, particularly when mass-producing silicon wafers, proper evaluation of the grinding wheel is also required. However, unlike silicon wafers, which are made of single-crystal silicon, grinding wheels are composed of a combination of abrasive grains and a binder, and chips and other debris are also present on their surfaces. Furthermore, while the surface of a silicon wafer has a very high degree of flatness, this structure results in a low degree of flatness. Due to this complex structure of grinding wheels, no standard has been established for non-destructive evaluation of grinding wheels. Furthermore, unlike silicon, which is generally in the form of a wafer, grinding wheels are fixed to a large base (metal base) and are used. Therefore, their evaluation form differs significantly from that of silicon wafers, making evaluation difficult.

[0006] The present invention has been made in view of the above problems, and an object of the present invention is to provide an invention that solves the above problems. [Means for solving the problem]

[0007] In order to solve the above problems, the present invention has the following configurations. The grinding wheel evaluation method of the present invention is a grinding wheel evaluation method for evaluating the working surface, which is the surface used in grinding, of a grinding wheel having a configuration in which abrasive grains made of diamond are dispersed in a binder and is used to grind a sample to be ground.The method comprises: a Raman spectrum acquisition step of acquiring, for each position on the working surface, a Raman spectrum obtained by scattered light from the working surface when monochromatic incident light is made incident on the working surface; and an analysis step of recognizing the peak wavenumber and half-width of the peak recognized from the Raman spectrum obtained for each position, and recognizing the abrasive grains, the binder, and the adhesion layer adhered to the working surface by grinding from the recognized peak wavenumber and half-width on the working surface. In the analyzing step, the abrasive grains may be identified from a peak in the Raman spectrum that corresponds to diamond. In the analysis step, damage to the abrasive grains or the adhesion layer may be recognized from a shift in the peak corresponding to diamond in the Raman spectrum. In the analyzing step, the adhesion layer containing the material constituting the sample to be ground may be identified from a peak corresponding to the material in the Raman spectrum. In the analysis step, the adhesion layer containing diamond may be recognized from an emission peak caused by defects in the diamond mixed into the Raman spectrum. The grinding wheel evaluation device of the present invention is a grinding wheel evaluation device that has a configuration in which abrasive grains are dispersed in a binder and that evaluates the working surface, which is the surface used in the grinding process, of a grinding wheel used in the grinding process.The grinding wheel evaluation device comprises: a light source that emits monochromatic incident light; a spectrometer that measures the Raman spectrum of light scattered by the working surface from the incident light; an optical system that directs the incident light to the working surface and directs the scattered light from the working surface to the spectrometer; a sample stage that adjusts the position of the grinding wheel so as to vary the incident position of the incident light on the working surface; and an analysis unit that adjusts the sample stage to obtain the Raman spectrum for each position on the working surface, recognizes the peak wavenumber and half-width of the peak recognized from the Raman spectrum obtained for each position, and recognizes the abrasive grains, the binder, and the adhesion layer that has adhered to the working surface by grinding from the recognized peak wavenumber and half-width. The optical system may include a deflection mirror that changes the optical paths of the incident light and the scattered light from a direction perpendicular to a surface of the sample stage to a direction parallel to the surface. [Effects of the Invention]

[0008] Since the present invention is configured as described above, it is possible to appropriately evaluate the grinding wheel working surface in a non-destructive manner. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a diagram showing a configuration of a grindstone evaluation device according to an embodiment of the present invention; [Figure 2]1 is a cross-sectional view schematically showing the structure of a grinding wheel to be evaluated in a grinding wheel evaluation according to an embodiment of the present invention. [Figure 3] 1 shows examples of Raman spectra obtained at a location with abrasive grains (a) and at a location with silicon chips (b) in the first embodiment. [Figure 4] 1 shows a structure other than the peaks attributable to diamond in the wave number range of 0 to 2000 cm −1 of the Raman spectrum in the first example. [Figure 5] This is an example of the first embodiment in which the Raman spectrum is contaminated with an emission peak caused by defects in the diamond. [Figure 6] 1 is an optical (microscopic) photograph of the grinding wheel surface measured in the first example. [Figure 7] In the first example, the peak intensities obtained from the Raman spectrum with wave numbers in the range of 1300 to 1400 cm-1 were mapped by focusing on the low intensity region (a) and the high intensity region (b). [Figure 8] In the first example, the wavenumber range is set to 1330 to 1343 cm-1 (a), the wavenumber range is set to 1331 to 1335 cm-1 (b), and the wavenumber range is set to 1343 to 1400 cm-1 (c), and the peak wavenumber positions are mapped. [Figure 9] In the first example, the peak widths are mapped in the range of 0 to 10 cm −1 (a), 10 to 40 cm −1 (b), and 40 to 50 cm −1 (c). [Figure 10] In the first example, the peak intensities obtained from the Raman spectrum with wave numbers in the range of 400 to 600 cm-1 were mapped by focusing on the low intensity region (a) and the high intensity region (b). [Figure 11] In the first example, the wavenumber range was set to 400 to 600 cm-1, and the peak wavenumber positions were mapped (a), and the pixels with low reliability due to the low peak intensity recognized in this result were masked (b). [Figure 12]In the first example, the wave number is in the range of 400 to 600 cm −1 , and the peak width is in the range of 1.6 to 12 cm −1 (a), and the peak width is in the range of 12 to 50 cm −1 (b). [Figure 13] 10 is an optical (microscopic) photograph of the grinding wheel surface measured in the second example. [Figure 14] In the second example, (a) shows the mapping result of the peak wavenumber position in the wavenumber range of 1331 to 1335 cm −1 , and (b) shows the mapping result of the peak width in the peak width range of 2 to 7 cm −1 . [Figure 15] In the second example, (a) shows the mapping result of the peak wavenumber position in the wavenumber range of 1335 to 1350 cm −1 , and (b) shows the mapping result of the peak width in the peak width range of 10 to 40 cm −1 . [Figure 16] In the second example, (a) shows the mapping result of the peak wavenumber position in the wavenumber range of 770 to 835 cm −1 , and (b) shows the mapping result of the peak width in the peak width range of 8 to 100 cm −1 . [Figure 17] 10 is an optical (microscopic) photograph of the grinding wheel surface measured in the third example. [Figure 18] In the third example, the Raman spectrum of the binder on the grinding wheel surface (a) and the sapphire sample to be ground (b) is shown. [Figure 19] 10 is a mapping result of the peak intensity at a wave number of 1330 cm −1 on the working surface of the grindstone in the third example. [Figure 20] 13 shows the results of mapping the peak wavenumber positions at 1332 to 1336 cm (a) and 1336 to 1340 cm (b) in the third example. [Figure 21] In the third example, the results are obtained by mapping peak widths in the ranges of 2 to 12 cm −1 (a) and 0 to 2 cm −1 (b). [Figure 22] This is the result of mapping the peak width in the range of 12 to 300 cm −1 in the third example. DETAILED DESCRIPTION OF THE INVENTION

[0010] A grinding wheel evaluation method and grinding wheel evaluation device according to an embodiment of the present invention will be described. This grinding wheel evaluation device (method) uses micro-Raman spectroscopy. FIG. 1 shows the configuration of this grinding wheel evaluation device 1. Here, micro-Raman spectroscopy is performed on the surface of a grinding wheel 100 (the working surface used for grinding: the surface on the left in FIG. 1). Monochromatic incident light S0 emitted from a light source 10 is imaged on the surface of a sample via a confocal optical system, and scattered light S1 from the sample is imaged on a spectroscope 16, thereby obtaining a Raman spectrum. In this case, the incident light S0 passes through a lens 11 and a pinhole 12, then passes through a beam splitter 13, and then through an objective lens 14 to be irradiated onto a sample stage 90.

[0011] The scattered light S1 emitted from the sample (on the sample stage 90 side) passes through the objective lens 14, is reflected vertically by the beam splitter 13, passes through the pinhole 15 and enters the spectroscope 16 equipped with a diffraction grating, where the spectrum of the scattered light S1 is measured. Here, in particular, the wave number (cm -1 ) is plotted on the horizontal axis (Raman shift) and the vertical axis is the detected intensity. The sample stage 90 is moved in the x' direction (left and right in the figure), y' direction (perpendicular to the paper), and z' direction (up and down in the figure) to map the Raman spectrum on the sample surface. Here, the beam splitter 13 and objective lens 14 function as an optical system that guides the incident light S0 toward the sample stage 90 and the scattered light S1 toward the spectroscope 16.

[0012] The above configuration is similar to that of a well-known Raman microscope (product name MR350, manufactured by SOL Instruments). The sample used here, the grinding wheel 100, is generally fixed at its bottom surface to a large base (metal base 200), and grinding is performed using its top surface. Therefore, if the working surface of the grinding wheel 100 in this configuration (fixed to the metal base 200) were to be made parallel to the surface (x'y' surface) of the sample stage 90 in FIG. 1 in order to perform a non-destructive evaluation of the grinding wheel 100, the sample (the structure consisting of the metal base 200 and the grinding wheel 100) fixed to the sample stage 90 would be very large in the z' direction in FIG. 1, making it difficult to place the sample on the sample stage 90.

[0013] For this reason, the sample is fixed to the sample stage 90 at a position away from the optical axis with the sample oriented horizontally (the stacking direction of the base metal 200 and grinding wheel 100 is horizontal), and a 2 mm high mirror (deflection mirror) 17 is installed on the optical axis to convert the optical path from the vertical direction to the horizontal direction in the figure (at an angle of 45° with the horizontal direction). The reflected light is then irradiated onto the grinding wheel 100. As a result, the optical path of the scattered light is also converted by 90°. In other words, in this case, the mirror (deflection mirror) 17 is added to the optical system to accommodate the particularly large sample (the grinding wheel 100 fixed to the base metal 200). As a result, data along the x, y, and z directions on the surface of the grinding wheel 100 in Figure 1 can be virtually obtained as data along the x', y', and z' directions.

[0014] Furthermore, the analysis unit 20, which is a computer, controls the sample stage 90 and acquires a Raman spectrum, as will be described later, at each point on the surface (working surface) of the grinding wheel 100 from the spectroscope 16. Thereafter, feature quantities, as will be described later, are extracted from the Raman spectrum, and the surface of the grinding wheel 100 is analyzed.

[0015] It is also possible to obtain an optical image of the sample by using a photodetector in place of the spectrometer 16 or by mapping the integrated intensity obtained by the spectrometer 16 .

[0016] Next, a schematic diagram of the cross-sectional structure near the surface of the grinding wheel 100 to be evaluated is shown in Figure 2. This grinding wheel 100 is formed by dispersing abrasive grains 110 made of diamond in a binder 120, and the lower surface in the figure is the working surface used for grinding. The abrasive grains 110 protruding from the surface of the binder 120 actually contribute to grinding.

[0017] Various defects in a grinding wheel 100 are shown schematically here. Typical defects include (A) shedding of abrasive grains 110 from the surface, (B) wear (fracture) of abrasive grains 110 protruding from the surface, and (C) clogging due to an adhesion layer 130 between the abrasive grains 110 protruding from the surface. The adhesion layer 130 is composed of materials (mainly silicon and its compounds) generated by grinding. Here, A can be recognized by identifying the abrasive grains 110, for example, by a decrease in their number. Therefore, when evaluating a grinding wheel 100, it is important to properly recognize the abrasive grains 110, the binder 120, and the adhesion layer 130 nondestructively. For the adhesion layer 130, it is important to recognize the substances contained therein. That is, this grinding wheel evaluation method includes a Raman spectrum acquisition process for acquiring Raman spectra for each position on the working surface of the grinding wheel 100 and an analysis process for analyzing the Raman spectra obtained for each position. In the following, it will be explained by way of examples that such recognition is possible particularly by evaluation of the Raman spectrum in this analysis step.

[0018] In the following, a laser light source with a wavelength of 532 nm and 5 mW was used as the light source 10, and an objective lens 14 with an NA of 0.5 and a magnification of 50 was used. The apertures of the pinholes 12 and 15 used for focusing were 30 μmφ. The diffraction grating in the spectroscope 16 was set to 600 lines / mm, so that the wavelength range of 0 to 3400 cm was obtained. -1 The data is 0.7cm -1 The scanning resolution (position resolution) of the sample stage 90 was set to 2 μm.

[0019] In the first example, the sample to be ground was silicon, the abrasive grains 110 of the grinding wheel 100 were #1000 (the grain size of the abrasive grains 110 was typically about 16 μm), and the binder 120 was resin-based. The object to be ground with the grinding wheel 100 was an 8-inch silicon wafer, the grinding fluid was water, and the amount of silicon removed by grinding was 200 μm.

[0020] Specific examples of the results of measuring the Raman spectrum of the grinding wheel 100 at each location after this grinding process are shown in Figures 3(a) and 3(b) along with the corresponding micrographs. Figure 3(a) corresponds to the position of the abrasive grain 110 in Figure 2. The prominent peak (1333 cm) observed here is -1 ) is due to the diamond crystals that make up the abrasive grains 110. On the other hand, FIG. 3(b) corresponds to a position other than the abrasive grains 110 in FIG. 2, and the prominent peak (520 cm -1 ) is thought to be caused by silicon (Si) crystals (the object being ground) and is a location where silicon chips are present.

[0021] On the other hand, the 0-2000 cm range in the Raman spectrum -1 Figure 4 shows an example of measuring the fine structure other than the above peaks. As can be seen, in addition to the large peaks mentioned above, many small peaks exist within this range. Among these, the peaks from 200 to 600 cm -1 The small peak at 740 cm is due to Si (Si-Si bond) or Si oxide (Si-O bond). -1 Relatively large peaks are observed at the surface of the film, which are due to the Si-C bond. These peaks correspond to the adhesion layer 130 (C in FIG. 2).

[0022] Also, 1000~1600cm -1 The small peak at 1333 cm is due to the C-C bond that constitutes the resin material, which corresponds to the binder 120 (resin). -1 ) is included in the wavenumber range of the peak (aggregate of small peaks) due to such C-C bonds, but the broadening (half width) of this peak in Figure 4 is larger than the broadening of the diamond peak in Figure 3(a).

[0023] Furthermore, the light detected by the spectrometer 16 is mainly light resulting from Raman scattering as described above, but if there is fluorescence emitted from the sample, this will also be detected at the same time. Figure 5 shows a spectrum in which such a fluorescence peak is mixed in addition to the peak in Figure 3(a). Here, the peak at 1400 cm -1 Near and 3100cm -1 The former is due to nitrogen (N) and neutral vacancies (V) in diamond. 0 ) bond, NV 0 The latter is due to the nitrogen (N) and electron-trapped vacancy (V - ) bond, NV - Both of these fluorescences are caused by diamond, but because of the large number of crystal defects, the abrasive grains 110 are damaged and nano-sized, resulting in a large number of crystal defects (corresponding to C (adhesion layer 130 or B) in Figure 2).

[0024] From the above results, by mapping the data extracted from the Raman spectra obtained at each position, the state shown in Figure 2 can be estimated. Here, as shown in Figure 1, an analyzer 20, which is a computer, is connected to the spectrometer 16. The analyzer 20 controls the sample stage 90 and recognizes the measurement results of the spectrometer 16 (the spectrum for each position), and can recognize the feature quantities of the spectrum. In this case, the feature quantities include the peak position (peak wavenumber), peak intensity, and peak width (half-width: peak width) when each peak is fitted with a Lorentz function. The results of mapping such feature quantities are explained below.

[0025] FIG. 6 is a micrograph of the measurement area of ​​the grinding wheel 100. On the other hand, FIG. 7 shows the measurement area of ​​the grinding wheel 100, which is measured using a 1300 to 1400 cm -1The mapping results for this measurement region are shown in Figure 7(a), where peak intensities in the range (the maximum intensity if there are multiple peaks) are used. Figure 7(a) shows the results for the low intensity range (0 to 300 cps: corresponding to Figure 3(b), where values ​​above 300 cps are displayed in black), and Figure 7(b) shows the results for the high intensity range (300 to 10,000 cps: corresponding to Figure 3(a), where values ​​below the lower limit are displayed in black and values ​​above the upper limit are displayed in white).

[0026] In particular, in Figure 7(b), based on the results of Figure 3, it is thought that the high-intensity region corresponds to the diamond abrasive grains 110, but in reality, a clear distinction cannot be made from this image. Similarly, a clear distinction cannot be made from Figure 6(b). In other words, in mapping using peak intensity as the above-mentioned feature, it is difficult to clearly distinguish between the abrasive grains 110 and the binder 120.

[0027] On the other hand, Figure 8 shows the mapping results using the peak wavenumber position as the above feature. Figure 8(a) shows the peak wavenumber position at 1330 to 1343 cm -1 Figure 8(b) shows a narrower wavenumber range of 1331 to 1335 cm -1 , and Fig. 8(c) shows the wavenumber range of 1343 to 1400 cm -1 The range of wavenumbers is shown in the figure, and the values ​​below the lower limit and above the upper limit are displayed in the same manner as above. Note that, since the wavenumber range in this case is narrow, there is only one peak in this wavenumber range.

[0028] Diamond peak at 1333cm -1 In the results of Figures 8(a) and (b), where the wavenumber range is included, the abrasive grain 110 can be clearly recognized. Furthermore, in Figure 8(b), where the wavenumber range is narrow, it can be confirmed that a distribution with continuous changes in peak position can be seen even within a single abrasive grain 110. Such changes in peak position within a single abrasive grain 110 are thought to correspond to the stress distribution in the abrasive grain 110 (diamond), and suggest that the abrasive grain 110 is damaged. In other words, if there is such a distribution in peak positions, or if the peak position is 1333 cm -1 The abrasive grain 110, which is recognized slightly shifted from the arrow B, corresponds to B in FIG.

[0029] Also, the diamond's 1333cm -1 In the results of FIG. 8(c), which is thought to not include the peak of NV in FIG. 5, the structure corresponding to the abrasive grain 110 is not visible, but other distributions are visible. 0 Since the central light emission is included, this distribution is considered to mainly reflect the adhesion layer 130 in FIG.

[0030] From the results of FIG. 3(a), the abrasive grains 110 (diamond) correspond to a narrow peak, and from the results of FIG. 4, the binder 120 (resin material) corresponds to a broad peak. Therefore, the above-mentioned feature amount is determined as 1300 to 1400 cm -1 The results of this case are shown in Fig. 9(a), where the peak width is 0 to 10 cm. -1 The range (narrowest case) is 10-40cm. -1 , and Fig. 9(c) is 40-50cm. -1 In either case, anything outside the above range is displayed as black.

[0031] 9(a), it is clear that the region with the narrowest peak width (sharpest peak) corresponds to the abrasive grain 110. Therefore, when the above-mentioned Raman spectrum is used, for example, the peak wave number is 1331 to 1335 cm -1 Between these, the peak width is 10cm -1 The following points can be assumed to be abrasive grains 110.

[0032] 9(b) and the distribution of the particles other than the abrasive grains 110 in FIG. 8(a) are similar, and the peak width in this range is wider than that of diamond crystals. Therefore, it is estimated that the distribution in FIG. 9(b) corresponds to the distribution of the binder 120. For example, the peak wave number is 1340 to 1345 cm -1 Between these, the peak width is 10 to 40 cm. -1 By mapping the range of the binder 12, the area corresponding to the binder 12 can be estimated.

[0033] The distribution in Figure 9(c) and the distribution in Figure 8(c) are common, and this distribution is the NV distribution as mentioned above. 0 It is believed to be related to the adhesion layer 130 (diamond layer with many defects) that contains the center.

[0034] Next, the peak of silicon (Fig. 3(b)) is detected in the 400-600 cm -1 The results of a similar analysis performed from the wave number range of 130 will be described below. Since silicon is contained only in the adhesion layer 130, this information is mainly related to the distribution of the adhesion layer 130 in FIG.

[0035] Figure 10 shows the above feature values, 400-600cm -1 The mapping results for this measurement region, which employed peak intensities in the range of 100 cps and 2000 cps, are shown. Figure 10(a) shows the results for low intensity (0 to 80 cps: 80 cps or higher is displayed in black), and Figure 10(b) shows the results for high intensity (80 to 2000 cps: 80 cps or lower is displayed in black, and 2000 cps or higher is displayed in white). From these results, no clear distribution is observed in the case of low intensity (Figure 10(a)), but in the case of high intensity (Figure 10(b)), a region with high peak intensity equivalent to the abrasive grains 110 can be confirmed. This region is thought to correspond to crystalline silicon, and is therefore presumed to be silicon chips during grinding, and corresponds to the adhesion layer 13 in Figure 2.

[0036] However, the peaks observed in this wavenumber region are not clear (the peak intensity is low), and as a result, the errors in the peak intensity, peak width, etc. obtained by fitting as described above are large. For this reason, the reliability of the distribution shown in Figure 10(a) is low.

[0037] Figure 11(a) shows the mapping results using the peak wavenumber in the wavenumber region as the feature. In this case, a distribution corresponding to crystalline silicon similar to that in Figure 10(b) is observed, but it is not clear. The lack of clarity is likely due to the lack of clarity of the recognized peaks, as in Figure 10(a). Therefore, Figure 11(b) shows the results of masking the pixels recognized as high intensity (low reliability pixels) in Figure 10(a) by displaying them in black. In Figure 11(b), the silicon chips are more clearly recognized.

[0038] Figure 12 shows the mapping results using the peak width in the above wavenumber range as the feature quantity. Figure 12(a) shows the peak widths of 1.6 to 12 cm. -1 In the range of , Fig. 12(b) shows that the peak width is 12 to 50 cm -1 12(a) shows the cases of crystalline silicon, which can be clearly recognized as a region with a narrow peak width. For example, the peak wavenumber is 511 to 529 cm -1 Peak width is 1.6 to 12 cm -1 Using the mapping of the range, silicon chips (corresponding to the adhesive layer 13 in FIG. 2) can be recognized.

[0039] As described above, the analysis unit 20 in Fig. 1 acquires Raman spectra for each position on the surface (working surface) of the grinding wheel 100 using the grinding wheel evaluation device 1 in Fig. 1, extracts feature quantities (peak intensity, peak wavenumber, and peak width in a specific wavenumber region) from the Raman spectra, and uses these to recognize the abrasive grains 110, binder 120, and adhesion layer 130 in Fig. 2. At this time, it is also possible to analyze damage to the abrasive grains 110 and the silicon component in the adhesion layer 130. In other words, this allows the condition of the working surface of the grinding wheel 100 to be recognized appropriately and non-destructively.

[0040] In the above example, the evaluation of a grinding wheel used to grind silicon crystals, which uses diamond as the abrasive grain and a resin material as the binder, was explained, but even if these materials are different, as long as the basic shape of the grinding wheel is the same as in Figure 2, it is possible to perform a similar evaluation by appropriately setting the peaks and fluorescence wavenumbers in the Raman spectrum. In this case, by using the device configured as in Figure 1, this evaluation can be performed non-destructively.

[0041] In this second example, the sample to be ground was silicon carbide (SiC), and the grinding wheel had the same structure as in the case where the sample to be ground was silicon. Figure 13 shows a micrograph of the grinding wheel working surface in this case (corresponding to Figure 6). Figure 14(a) shows the 1331-1335 cm corresponding to diamond. -1 The results of mapping the peak wavenumber positions in the wavenumber range of 2 to 7 cm (corresponding to Figure 8(b)). Figure 14(b) shows the peak wavenumber positions in the wavenumber range of 2 to 7 cm. -1 9(a) shows the results of mapping the peak width in the range of 1.0 mm. In this result, too, the abrasive grains (diamond) can be recognized, just like in the first example.

[0042] Similarly, Figure 15(a) shows the peak at 1335-1350 cm corresponding to the binder. -1 The results of mapping the peak wavenumber positions in the wavenumber range of 10 to 40 cm (corresponding to Fig. 8(c)). -1 9(b) shows the results of mapping the peak width in the range of 1.0 to 1.0 mm. In this result, too, the binder (resin-based material) can be recognized, as in the first example.

[0043] Similarly, Fig. 16(a) shows the bands of 770 to 835 cm corresponding to the Si-C bond of the ground sample. -1 The results of mapping the peak wavenumber positions in the wavenumber range of 8 to 100 cm (corresponding to Figure 11). Figure 16(b) shows the peak wavenumber positions in the wavenumber range of 8 to 100 cm. -1 This shows the results of mapping the peak width in the range (corresponding to Figure 12). This makes the SiC chips (adhesion layer) visible.

[0044] Next, in the third example, the grinding sample was sapphire (corundum-type aluminum oxide single crystal), and the binder in the grinding wheel was an inorganic glassy material (vitrified binder). Figure 17 is a micrograph of the sample, and Figure 18(a) shows the Raman spectrum of the binder area in the photo. Here, in the vitrified binder, two separate phases are formed, but the peak widths of the peaks corresponding to both phases are wide. On the other hand, Figure 18(b) shows the Raman spectrum of the sapphire grinding sample. Here, the 417 cm -1 Although a sharp peak is observed around 1000 nm, the scattering intensity itself is not high, so it is not easy to identify the sapphire chips in this example.

[0045] Figure 19 shows the wave number of 1330 cm -1 This is a mapping result of the peak intensity of 1333 to 1336 cm, which corresponds to the distribution of abrasive grains (diamonds). -1 8(b), a continuous distribution of peak wavenumbers is observed in this figure, which corresponds to the residual stress distribution in the abrasive grains. Also, Figure 20(b) shows the wavenumber range of 1336-1340 cm adjacent to the above wavenumber range in the same region. -1 20(a) is not visible, but instead, a region with a peak within this wavenumber range is seen. -1 The region with a peak at 1336–1340 cm is presumed to be abrasive grains exposed on the surface, and the peak at 1336–1340 cm is recognized in Figure 20(b). -1 The region with a peak in is thought to correspond to the abrasive grains embedded inside the binder, and this is thought to indicate that the abrasive grains embedded inside the binder were subjected to particularly large compressive stress due to the pressure applied during the formation of the grinding wheel, resulting in a shift in the peak wavenumber.

[0046] Figure 21(a) shows the peak width in the same region from 2 to 12 cm. -1Figure 21(b) shows the results of mapping in the range of 0 to 2 cm. -1 The area recognized in Figure 21(a) is the same as the area recognized in Figure 20(a) (exposed abrasive grains), and the area recognized in Figure 21(b) is the same as the area recognized in Figure 20(b) (buried abrasive grains). The wider peak width in the area seen in Figure 21(a) is thought to correspond to the exposed abrasive grains being damaged.

[0047] Figure 22 shows the peak width from 12 to 300 cm -1 This is the result of mapping in the range of 19 to 21. The results show a distribution that is completely different from those in Figures 19 to 21. This distribution has a particularly wide peak width, which is thought to reflect diamond-like carbon, etc., which has an amorphous structure at the tip (surface) of the abrasive grains and in the adhesion layer.

[0048] In this way, by setting the range of peak wavenumbers and the range of peak half-width depending on the binder and the material of the sample to be ground, the condition of the working surface of the grinding wheel can be appropriately evaluated.

[0049] The present invention has been described above based on the embodiments. However, these embodiments are merely examples, and various modifications are possible in the combination of the respective components. It will be understood by those skilled in the art that such modifications are also within the scope of the present invention. [Explanation of symbols]

[0050] 1. Grinding wheel evaluation device 10 light source 11 Lens 12, 15 pinhole 13 Beam splitter 14 Objective Lens 16 Spectrometer 17 Mirror (deflecting mirror) 20 Analysis Department 90 Sample stage 100 whetstones 110 abrasive grains 120 Binder 130 Adhesion layer 200 base money S0 incident light S1 Scattered light

Claims

1. A grinding wheel evaluation method for evaluating a working surface, which is a surface used in grinding, of a grinding wheel having a structure in which abrasive grains made of diamond are dispersed in a binder and is used in grinding a sample to be ground, comprising: a Raman spectrum acquisition step of acquiring a Raman spectrum obtained by scattering light from the work surface when monochromatic incident light is incident on the work surface for each position on the work surface; an analysis step of identifying the peak wavenumber and half-width of the peak recognized from the Raman spectrum obtained for each position, and identifying the abrasive grains, the binder, and the adhesion layer adhered to the working surface by grinding on the basis of the recognized peak wavenumber and half-width; A grinding wheel evaluation method comprising:

2. In the analyzing step, 2. The grindstone evaluation method according to claim 1, wherein the abrasive grains are identified from a peak in the Raman spectrum that corresponds to diamond.

3. In the analyzing step, 3. The grinding wheel evaluation method according to claim 2, wherein damage to the abrasive grains or the adhesion layer is recognized from a shift in the peak corresponding to diamond in the Raman spectrum.

4. In the analyzing step, 3. The grinding wheel evaluation method according to claim 1, wherein the adhesion layer containing the material is recognized from a peak in the Raman spectrum corresponding to the material constituting the grinding sample.

5. In the analyzing step, 3. The grinding wheel evaluation method according to claim 1, wherein the adhesion layer containing diamond is recognized from an emission peak caused by defects in the diamond mixed in the Raman spectrum.

6. A grinding wheel evaluation device for evaluating a working surface of a grinding wheel used in a grinding process, the working surface being a surface used in the grinding process, the working surface having a structure in which abrasive grains are dispersed in a binder, comprising: a light source that emits monochromatic incident light; a spectrometer for measuring a Raman spectrum of scattered light from the work surface; an optical system that guides the incident light to the work surface and guides the scattered light from the work surface to the spectroscope; a sample stage for adjusting the position of the grindstone so as to vary the incident position of the incident light on the work surface; an analysis unit that acquires the Raman spectrum for each position on the working surface by adjusting the sample stage, recognizes a peak wavenumber and a half-width of a peak recognized from the Raman spectrum obtained for each position, and recognizes the abrasive grains, the binder, and an adhesion layer adhered to the working surface by grinding on the basis of the recognized peak wavenumber and the half-width; A grinding wheel evaluation device comprising:

7. 7. The grindstone evaluation device according to claim 6, wherein the optical system includes a deflection mirror that changes the optical paths of the incident light and the scattered light from a direction perpendicular to the surface of the sample stage to a direction parallel to the surface.

Images