Pad surface determination method and pad surface determination system

Abstract

The present invention relates to a pad surface determination method for determining the surface properties of a polishing pad for polishing a wafer or other substrate. The pad surface determination method comprises simultaneously irradiating a target area (T) within a polishing surface (2a) with a plurality of light of different wavelengths from different directions when viewed from a direction perpendicular to the polishing surface (2a), receiving a plurality of reflected light from the target area (T) at an imaging device (59), generating, by means of the imaging device (59), a plurality of images from the plurality of reflected light, each image corresponding to a respective one of the plurality of light of different wavelengths, generating an evaluation image, which is a difference image or a division image of the plurality of images, and determining a surface property of the polishing pad (2) on the basis of luminance information in the evaluation image.

Description

Pad surface determination method and pad surface determination system

[0001] The present invention relates to a pad surface evaluation method and a pad surface evaluation system for evaluating the surface properties of a polishing pad used to polish a substrate such as a wafer.

[0002] In the manufacturing process of semiconductor devices, planarization of the semiconductor device surface is becoming increasingly important. The most important technology for this surface planarization is chemical mechanical polishing (CMP). Chemical mechanical polishing (hereinafter referred to as CMP) is a process for polishing a surface of a semiconductor device using silica (SiO ２ This is a process in which a polishing liquid containing abrasive grains such as ammonium hydroxide is supplied onto the polishing surface of a polishing pad while the substrate, such as a wafer, is brought into sliding contact with the polishing surface to polish the substrate.

[0003] A polishing apparatus for performing CMP includes a polishing table that supports a polishing pad having a polishing surface, and a polishing head that presses a substrate against the polishing pad. The polishing apparatus polishes a substrate as follows: A polishing liquid (typically a slurry) is supplied to the polishing surface of the polishing pad while the polishing table and polishing pad are rotated together. The polishing head presses the surface of the substrate against the polishing surface of the polishing pad while rotating the substrate. The substrate is brought into sliding contact with the polishing pad in the presence of the polishing liquid. The surface of the substrate is polished by the chemical action of the polishing liquid and the mechanical action of the abrasive grains contained in the polishing liquid and / or the polishing pad.

[0004] When polishing a substrate, abrasive grains and polishing debris adhere to the polishing surface of the polishing pad, reducing its polishing performance. Therefore, to restore the polishing surface of the polishing pad, dressing (conditioning) of the polishing pad is performed using a dresser. The dresser has hard abrasive grains, such as diamond particles, fixed to its underside, and the polishing surface of the polishing pad is regenerated by scraping it off with the dresser.

[0005] JP 2014-172153 A International Publication No. 2005 / 072910 JP 2022-112194 A

[0006] The polishing rate (also called removal rate) of a substrate can vary depending on the condition of the polishing surface of a polishing pad. A polishing pad has recesses, such as grooves or holes, formed on its polishing surface to hold the polishing liquid. As the polishing pad wears, the depth of the recesses decreases, making it difficult for the polishing pad to hold the polishing liquid on its polishing surface. Furthermore, as multiple substrates are polished on the polishing pad, polishing debris gradually accumulates in the recesses. When the recesses are filled with polishing debris, it becomes difficult for the polishing pad to hold the polishing liquid on its polishing surface. As a result, the polishing rate of the substrate decreases.

[0007] The polishing surface of a polishing pad has minute protrusions in addition to the recesses described above. These protrusions form the surface roughness of the polishing pad. If the protrusions on the polishing surface become smaller (i.e., if the surface roughness of the polishing pad becomes lower), the removal rate of the substrate may decrease.

[0008] Therefore, the present invention provides a pad surface evaluation method and a pad surface evaluation system for evaluating the surface properties of a polishing pad, including the state of the concave and convex portions of the polishing surface of the polishing pad.

[0009] In one aspect, a pad surface evaluation method is provided for evaluating the surface properties of a polishing pad having a polishing surface for polishing a substrate, the method comprising: simultaneously irradiating a target area within the polishing surface with multiple lights of different wavelengths from different directions when viewed from a direction perpendicular to the polishing surface; receiving multiple reflected lights from the target area with an imaging device; generating multiple images from the multiple reflected lights corresponding to the multiple lights of different wavelengths with the imaging device; generating an evaluation image which is either a difference image or a division image of the multiple images; and evaluating the surface properties of the polishing pad based on brightness information of the evaluation image.

[0010] In one aspect, the angles of incidence of the plurality of light beams having different wavelengths on the target area are within a range of 3 to 80 degrees. In one aspect, the angles of incidence of the plurality of light beams having different wavelengths on the target area are the same. In one aspect, the plurality of images are generated simultaneously by the imaging device. In one aspect, determining the surface quality of the polishing pad based on luminance information of the evaluation image comprises calculating a pad surface index value representing the quality of the polishing surface based on the luminance information of the evaluation image. In one aspect, the pad surface evaluation method further includes repeatedly irradiating the target area with the plurality of light beams having different wavelengths, generating a plurality of images from the plurality of reflected light beams by the imaging device, generating the evaluation image from the plurality of images, and calculating the pad surface index value based on luminance information of the evaluation image, thereby acquiring a plurality of pad surface index values ​​and calculating a variance of the plurality of pad surface index values.

[0011] In one aspect, the step of calculating the pad surface index value based on the luminance information of the evaluation image is a step of calculating the pad surface index value using luminance values ​​within an acceptable range among the luminance values ​​of pixels constituting the evaluation image. In one aspect, the pad surface determination method further includes creating a frequency distribution of the luminance values ​​of the pixels constituting the evaluation image, calculating a standard deviation from the frequency distribution, and determining the acceptable range from the standard deviation. In one aspect, the step of generating the evaluation image from the plurality of images is a step of correcting at least one image of the plurality of images by removing luminance values ​​outside an acceptable range from the luminance values ​​of pixels constituting the at least one image, and generating the evaluation image, which is either a difference image or a division image, from the plurality of images including the corrected image. In one aspect, the pad surface determination method further includes creating a frequency distribution of the luminance values ​​of the pixels constituting the at least one image, calculating a standard deviation from the frequency distribution, and determining the acceptable range from the standard deviation. In one aspect, judging the surface quality of the polishing pad based on the brightness information of the evaluation image involves inputting the brightness information of the evaluation image into a judgment model constructed by machine learning, and outputting a judgment result of the surface quality of the polishing pad from the judgment model.

[0012] In one aspect, a pad surface evaluation system is provided for evaluating the surface properties of a polishing pad having a polishing surface for polishing a substrate, the pad surface evaluation system comprising: a plurality of light projecting units that simultaneously irradiate a target area within the polishing surface with a plurality of light beams of different wavelengths from different directions when viewed from a direction perpendicular to the polishing surface; an imaging device that receives a plurality of reflected light beams from the target area and generates a plurality of images from the plurality of reflected light beams corresponding to the plurality of light beams of different wavelengths; and a system processing unit that generates an evaluation image, which is either a difference image or a division image, from the plurality of images and evaluates the surface properties of the polishing pad based on the brightness information of the evaluation image.

[0013] In one aspect, the angles of incidence of the plurality of light beams having different wavelengths on the target area are within a range of 3 to 80 degrees. In one aspect, the angles of incidence of the plurality of light beams having different wavelengths on the target area are the same. In one aspect, the imaging device is configured to simultaneously generate the plurality of images. In one aspect, the system processing unit is configured to calculate a pad surface index value representing the properties of the polishing surface based on brightness information of the evaluation image. In one aspect, the system processing unit is configured to calculate the variance of the plurality of pad surface index values ​​calculated at different times. In one aspect, the system processing unit is configured to calculate the pad surface index value using brightness values ​​within an acceptable range of brightness values ​​of pixels constituting the evaluation image. In one aspect, the system processing unit is configured to create a frequency distribution of brightness values ​​of pixels constituting the evaluation image, calculate a standard deviation from the frequency distribution, and determine the acceptable range from the standard deviation. In one aspect, the system processing unit is configured to correct at least one of the multiple images by removing luminance values ​​outside an acceptable range from the luminance values ​​of pixels constituting at least one of the multiple images, and generate the evaluation image, which is either a difference image or a division image, from the multiple images including the corrected image. In one aspect, the system processing unit is configured to create a frequency distribution of the luminance values ​​of the pixels constituting the at least one image, calculate a standard deviation from the frequency distribution, and determine the acceptable range from the standard deviation. In one aspect, the system processing unit is configured to input luminance information of the evaluation image to a determination model constructed by machine learning, and output a determination result of the surface texture of the polishing pad from the determination model.

[0014] Multiple beams of light are irradiated onto the same target area from different directions when viewed perpendicular to the polishing surface. Multiple images corresponding to the multiple beams of light incident on the target area are generated from multiple beams of light reflected from the polishing surface, and an evaluation image is further generated from the multiple images. The intensity of the multiple beams of light reflected varies according to the depth of the recesses and the height of the protrusions on the polishing surface. The evaluation image exhibits characteristic changes according to changes in the surface texture of the polishing pad. Therefore, it is possible to accurately determine the surface condition of the polishing pad based on the brightness information of the evaluation image.

[0015] 1 is a top view showing an embodiment of a polishing apparatus; FIG. 2 is a side view of the polishing apparatus shown in FIG. 1; FIG. 3A is a diagram showing an example of a groove pattern formed on the polishing surface of a polishing pad; FIG. 3B is a diagram showing an example of a groove pattern formed on the polishing surface of a polishing pad; FIG. 3C is a diagram showing an example of a groove pattern formed on the polishing surface of a polishing pad; FIG. 3D is a diagram showing an example of a groove pattern formed on the polishing surface of a polishing pad; FIG. 3D is a diagram showing an example of a hole formed on the polishing surface of a polishing pad; and FIG. 6A is a diagram explaining how the intensity of reflected light from the recesses changes as the polishing pad wears. FIG. 6B is a diagram explaining how the intensity of reflected light from the recesses changes as the polishing pad wears. FIG. 7A is a diagram explaining how the intensity of reflected light from the recesses changes as polishing debris accumulates in the recesses of the polishing pad. FIG. 7B is a diagram explaining how the intensity of reflected light from the recesses changes as polishing debris accumulates in the recesses of the polishing pad. Schematic diagrams showing examples of images in which brightness changes according to changes in the depth of the recesses. 9A is a diagram illustrating how the intensity of reflected light changes according to changes in the height of convex portions formed on the polishing surface of a polishing pad. FIG. 9B ...C is a schematic diagram illustrating an example of an image in which brightness changes according to changes in the height of convex portions. FIG. 9D is a diagram illustrating an example of generating two images from reflected light corresponding to light emitted from two light-projecting units and creating an evaluation image from these images. FIG. 9E is a diagram illustrating another example of generating two images from reflected light corresponding to light emitted from two light-projecting units and creating an evaluation image from these images. FIG. 9F is a schematic diagram illustrating an embodiment of a pad surface determination system having three light-projecting units. FIG. 9G is a diagram illustrating an embodiment in which light is irradiated from three light-projecting units onto a target area containing convex portions. FIG. 9H is a diagram illustrating an example of a plurality of segment areas set within an image. FIG. 9I is a diagram illustrating an example of a first area and a second area set within an image. FIG. 9I is a graph illustrating an example of a pad surface index value that changes with the usage time of a polishing pad. FIG. 9I is a schematic diagram illustrating an example of black spots and a white pattern on an image. FIG. 9I is a diagram illustrating an example of a frequency distribution.1 is a diagram showing another embodiment of a pad surface determination system; FIG. 2 is a schematic diagram showing an example of a determination model constructed using a deep learning method;

[0016] Embodiments of the present invention will now be described with reference to the drawings. Fig. 1 is a top view showing one embodiment of a polishing apparatus. Fig. 2 is a side view of the polishing apparatus shown in Fig. 1. The polishing apparatus chemically and mechanically polishes a substrate W, such as a wafer. As shown in Figs. 1 and 2, this polishing apparatus includes a polishing table 3 supporting a polishing pad 2 having a polishing surface 2a, a polishing head 1 pressing the substrate W against the polishing surface 2a, a polishing liquid supply nozzle 5 supplying a polishing liquid (e.g., a slurry containing abrasive grains) to the polishing surface 2a, and a pad surface evaluation system 40 for evaluating the properties of the polishing surface 2a of the polishing pad 2.

[0017] The polishing apparatus further includes a support shaft 14, a polishing head swing arm 16 connected to the upper end of the support shaft 14, and a polishing head shaft 10 rotatably supported at the free end of the polishing head swing arm 16. The polishing head 1 is fixed to the lower end of the polishing head shaft 10. The polishing head 1 is configured to be able to hold a substrate W on its lower surface. The substrate W is held with the surface to be polished facing downward.

[0018] A polishing head rotating mechanism (not shown) equipped with an electric motor or the like is disposed inside the polishing head swing arm 16. This polishing head rotating mechanism is connected to the polishing head shaft 10 and is configured to rotate the polishing head shaft 10 and the polishing head 1 around the axis of the polishing head shaft 10.

[0019] The polishing head shaft 10 is connected to a polishing head lifting mechanism (including, for example, a ball screw mechanism) not shown. This polishing head lifting mechanism is configured to move the polishing head shaft 10 up and down relative to the polishing head swing arm 16. The up and down movement of the polishing head shaft 10 allows the polishing head 1 to move up and down relative to the polishing head swing arm 16 and the polishing table 3.

[0020] The polishing apparatus further includes a table motor 6 that rotates the polishing table 3 together with the polishing pad 2. The table motor 6 is disposed below the polishing table 3, and the polishing table 3 is connected to the table motor 6 via a table shaft 3a. The polishing table 3 and the polishing pad 2 are rotated around the axis of the table shaft 3a by the table motor 6. The polishing pad 2 is affixed to the upper surface of the polishing table 3. The exposed surface of the polishing pad 2 forms a polishing surface 2a that polishes a substrate W such as a wafer.

[0021] The polishing apparatus further includes a polishing control unit 50 that controls its operation. The polishing head 1, polishing head rotation mechanism, polishing head lifting mechanism, polishing liquid supply nozzle 5, table motor 6, and pad surface determination system 40 are electrically connected to the polishing control unit 50, and the operations of the polishing head 1, polishing head rotation mechanism, polishing head lifting mechanism, polishing liquid supply nozzle 5, table motor 6, and pad surface determination system 40 are controlled by the polishing control unit 50.

[0022] The dressing control unit 50 is composed of at least one computer. The dressing control unit 50 includes a storage device 50a that stores a program for controlling the operation of the dressing machine, and an arithmetic unit 50b that executes calculations according to instructions included in the program. The storage device 50a includes a main storage device such as a 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 arithmetic unit 50b include a CPU (central processing unit) and a GPU (graphics processing unit). However, the specific configuration of the dressing control unit 50 is not limited to these examples.

[0023] The substrate W is polished as follows. While the polishing table 3 and polishing head 1 are rotated in the directions indicated by the arrows in Figures 1 and 2, a polishing liquid is supplied from a polishing liquid supply nozzle 5 to the polishing surface 2a of the polishing pad 2 on the polishing table 3. While the substrate W is rotated by the polishing head 1, the polishing head 1 presses the substrate W against the polishing surface 2a of the polishing pad 2 with the polishing liquid present on the polishing pad 2. The surface of the substrate W is polished by the chemical action of the polishing liquid and the mechanical action of the abrasive grains contained in the polishing liquid and / or the polishing pad 2. Thereafter, the substrate W may be water-polished while pure water is supplied onto the polishing pad 2 from a pure water nozzle (not shown).

[0024] After polishing of the substrate W is completed, the substrate W is moved outside the polishing pad 2 and transported to a device for the next processing. Then, a dresser (not shown) is used to dress the polishing surface 2 a of the polishing pad 2. The dresser has hard abrasive grains such as diamond particles fixed to its underside, and the polishing surface 2 a of the polishing pad 2 is slightly scraped away by this dresser, thereby regenerating the polishing surface 2 a of the polishing pad 2.

[0025] Grooves and / or holes (bottomed holes) having a predetermined pattern are formed on the polishing surface 2a of the polishing pad 2. Figures 3A to 3D show examples of groove patterns formed on the polishing surface 2a of the polishing pad 2. Figure 3A shows a lattice pattern, Figure 3B shows a radial pattern, Figure 3C shows a concentric pattern, and Figure 3D shows a spiral pattern. Figure 4 shows an example of holes formed on the polishing surface 2a of the polishing pad 2. The holes are formed over the entire polishing surface 2a of the polishing pad 2. The polishing surface 2a of the polishing pad 2 may have both grooves and holes, or either grooves or holes. The grooves and holes are formed for the purpose of, for example, uniformly distributing the polishing liquid over the entire substrate W. In this specification, the grooves and holes formed in the polishing pad 2 are collectively referred to as "recesses."

[0026] The polishing surface 2a of the polishing pad 2 gradually wears away with repeated polishing of the substrate W and dressing of the polishing pad 2. As a result, the depth of the recesses (holes, grooves) decreases, making it difficult for the polishing pad 2 to retain polishing liquid on its polishing surface 2a. Furthermore, polishing debris and the like become clogged in the recesses formed on the polishing surface 2a. Such changes in the surface texture of the polishing pad 2 degrade the polishing performance of the polishing pad 2, ultimately reducing the polishing rate when polishing the substrate W. In addition to the recesses described above, the polishing surface 2a of the polishing pad 2 also has minute protrusions. These protrusions form the surface roughness of the polishing pad 2. If the protrusions on the polishing surface 2a become smaller (i.e., if the surface roughness of the polishing pad 2 decreases), the polishing rate of the substrate W may decrease. Therefore, the polishing apparatus of this embodiment is equipped with a pad surface evaluation system 40 that evaluates the surface texture of the polishing pad 2. The surface texture of the polishing pad 2 includes the state of the recesses and protrusions on the polishing surface 2a. The convex portions of the polishing surface 2 a may include not only the convex portions of the polishing surface 2 a itself, but also foreign matter (abrasive grains, polishing debris, etc.) on the polishing surface 2 a. Examples of the convex portions of the polishing surface 2 a itself include both protrusions that are the initial surface shape of the polishing pad 2 formed during the manufacture of the polishing pad 2, and minute protrusions formed by diamond particles of a dresser during dressing (conditioning) of the polishing pad 2.

[0027] The pad surface evaluation system 40 includes a plurality of light projecting units 51, 52 that simultaneously irradiate a target area T on the polishing surface 2a with a plurality of light beams of different wavelengths from different directions, an imaging device 59 that receives a plurality of light beams reflected from the target area T and generates a plurality of images from the plurality of reflected light beams corresponding to the plurality of light beams of different wavelengths, and a system processing unit 70 that generates an evaluation image, which is either a difference image or a division image, from the plurality of images and evaluates the surface texture of the polishing pad 2 based on brightness information of the evaluation image. As will be described later, the system processing unit 70 is configured to calculate a pad surface index value that represents the texture of the polishing surface 2a based on the brightness information of the evaluation image, and to evaluate the surface texture of the polishing pad 2 based on the pad surface index value.

[0028] The system processing unit 70 is composed of at least one computer. The system processing unit 70 includes a storage device 70a that stores a program for determining the properties of the polishing surface 2a of the polishing pad 2, and an arithmetic unit 70b that executes calculations according to instructions included in the program. The storage device 70a includes a main storage device such as a 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 arithmetic unit 70b include a CPU (central processing unit) and a GPU (graphics processing unit). However, the specific configuration of the system processing unit 70 is not limited to these examples.

[0029] The light projecting units 51, 52 and the image capturing device 59 are disposed above the polishing surface 2a of the polishing pad 2. The light projecting units 51, 52 and the image capturing device 59 are fixed to a support member (not shown), and the relative positions of the light projecting units 51, 52 and the image capturing device 59 with respect to the polishing table 3 and the polishing pad 2 are fixed. The system processing unit 70 is electrically connected to the light projecting units 51, 52 and the image capturing device 59, and the operations of the light projecting units 51, 52 are controlled by the system processing unit 70.

[0030] The light projecting units 51, 52 include light sources such as lasers, light-emitting diodes (LEDs), and strobe flash light sources (e.g., xenon flash lamps). In addition to the light source, each of the light projecting units 51, 52 may further include optical elements such as lenses and fiber optic cables that direct light generated from the light sources to the target area T within the polishing surface 2a. In this embodiment, the multiple light projecting units 51, 52 include a first light projecting unit 51 and a second light projecting unit 52. In one embodiment, the light projecting units 51, 52 are installed at the same elevation angle with respect to the polishing surface 2a of the polishing pad 2. In another embodiment, an additional light projecting unit may be installed at an elevation angle different from that of the light projecting units 51, 52 with respect to the polishing surface 2a.

[0031] 1, the light projecting units 51 and 52 are configured to project light onto the target area T from different directions when viewed from a direction perpendicular to the polishing surface 2a. The angle α between the light emission directions of the light projecting units 51 and 52 when viewed from a direction perpendicular to the polishing surface 2a is within a range of 10 to 170 degrees, preferably 30 to 150 degrees, and more preferably 60 to 120 degrees. In one embodiment, three or more light projecting units may be provided as long as they can project light onto the target area T from different directions.

[0032] The system processing unit 70 issues commands to the multiple light-projecting units 51, 52 to cause the light-projecting units 51, 52 to emit light simultaneously. Therefore, the light from the light-projecting units 51, 52 overlaps within the target area T. As long as the light from the multiple light-projecting units 51, 52 is irradiated simultaneously onto the target area T, the light emission start times and light emission stop times of the multiple light-projecting units 51, 52 may be different. The target area T is the area where the light emitted from the multiple light-projecting units 51, 52 is incident on the polishing surface 2a.

[0033] In one embodiment, the system processing unit 70 may be configured to issue commands to the first light-projecting unit 51 and the second light-projecting unit 52 to constantly generate first light and second light having different wavelengths, while issuing commands to the imaging device 59 to simultaneously generate a first image and a second image from first reflected light and second reflected light corresponding to the first light and second light, respectively.

[0034] The arrangement pitch of the recesses (grooves and / or blind holes) is smaller than the width of the target area T. Therefore, at least one recess as described with reference to FIGS. 3A to 3D and 4 exists within the target area T. Therefore, the light emitted from each of the light-projecting units 51 and 52 is incident on the recess within the target area T. Similarly, at least one protrusion exists within the target area T. Therefore, the light emitted from each of the light-projecting units 51 and 52 is incident on the protrusion within the target area T.

[0035] The imaging device 59 includes an image sensor 58 that generates a color image including red, green, and blue. The imaging device 59 is disposed in a position where it can receive light reflected from the target area T, which includes concave and convex portions. The imaging device 59 may further include an optical element such as a lens. Examples of the image sensor 58 include a CCD sensor and a CMOS sensor. The multiple light projecting units 51, 52 and the imaging device 59 face the target area T. The imaging device 59 is disposed above the target area T. The imaging device 59 generates multiple images from the multiple reflected lights from the target area T and transmits these images to the system processing unit 70.

[0036] In this embodiment, the imaging device 59 including the image sensor 58 is configured to simultaneously generate a first image and a second image from first reflected light and second reflected light corresponding to the first light and second light emitted from the light projecting units 51 and 52, respectively. The configuration of the imaging device 59 that generates two images from the superimposed first reflected light and second reflected light is not particularly limited.

[0037] In one embodiment, the imaging device 59 may include a prism that separates the superimposed first reflected light and second reflected light into the first reflected light and the second reflected light, and two image sensors that receive the separated first reflected light and second reflected light. In another embodiment, the imaging device 59 may include a color filter that separately extracts the first reflected light and the second reflected light from the superimposed first reflected light and the second reflected light, and an image sensor that receives the first reflected light and the second reflected light that have passed through the color filter. In yet another embodiment, the imaging device 59 may include an image sensor having two groups of pixels that respectively receive the superimposed first reflected light and the second reflected light. The imaging device 59 can selectively receive light of each wavelength to generate an image. Therefore, the imaging device 59 can simultaneously generate multiple images, thereby reducing the time required to generate a pad surface index value. Since the imaging device 59 can generate multiple images simultaneously, it is possible to reduce the positional deviation of each image even when the polishing surface 2 is in a dynamic state, and it is also possible to shorten the time required to generate a pad surface index value.

[0038] 5 is a cross-sectional view showing an example of a recess formed on the polishing surface 2a of a brand new polishing pad 2 that has not been used to polish a substrate. The aspect ratio of the recess 75 is expressed as Lb / La, where La is the width of the recess 75 and Lb is the depth of the recess 75. The incident angle θ of the light emitted from the light-emitting units 51 and 52 is defined as the angle of the light with respect to the polishing surface 2a. The incident angle θ of the light emitted from the light-emitting units 51 and 52 is smaller than the angle determined by the aspect ratio of the recess 75. More specifically, the incident angle θ of the multiple light beams emitted from the light-emitting units 51 and 52 satisfies the condition tan θ<Lb / La (the aspect ratio of the recess 75).

[0039] 6A and 6B are diagrams illustrating how the intensity of reflected light from recesses 75 changes as polishing pad 2 wears. As shown in Fig. 6A, when polishing pad 2 is not worn, light with an incident angle θ strikes the side surface of recesses 75, resulting in a low intensity of reflected light from recesses 75. As shown in Fig. 6B, as polishing pad 2 wears, light with an incident angle θ strikes the bottom surface of recesses 75, resulting in an increase in the intensity of reflected light from recesses 75.

[0040] 7A and 7B are diagrams illustrating how the intensity of reflected light from recess 75 changes as polishing debris accumulates in recess 75 of polishing pad 2. As shown in Fig. 7A, when polishing debris does not accumulate in recess 75 of polishing pad 2, light with an incident angle θ strikes the side surface of recess 75, resulting in a low intensity of reflected light from recess 75. As shown in Fig. 7B, when polishing debris accumulates in recess 75, light with an incident angle θ strikes the accumulated polishing debris, resulting in an increase in the intensity of reflected light from recess 75.

[0041] In this way, the intensity of the reflected light from the target area T including the recess 75 changes with wear of the polishing pad 2 and accumulation of polishing debris in the recess 75, that is, with changes in the properties of the polishing surface 2a of the polishing pad 2.

[0042] 8 is a schematic diagram showing an example of an image in which the brightness changes according to the change in the depth of the recess. In the example shown in Fig. 8, the light projecting units 51 and 52 emit light at an incident angle of 30°. The low-brightness areas in each image in Fig. 8 correspond to the shadows of the recesses 75 on the polishing surface 2a of the polishing pad 2.

[0043] 8, the brightness of the image changes as the depth of the recess 75 changes. In particular, the point at which the brightness of the image changes significantly depends on the angle of incidence of light and the aspect ratio of the recess 75. Although not shown, for the same reason, the brightness of the image also changes depending on the amount of polishing debris accumulated in the recess 75 of the polishing pad 2.

[0044] 9A and 9B are diagrams illustrating how the intensity of reflected light changes according to the change in height of the convex portions 76 formed on the polishing surface 2a of the polishing pad 2. When light is incident on the polishing surface 2a of the polishing pad 2 at an incident angle θ, part of the light is blocked by the convex portions 76. As a result, the shadows of the convex portions 76 are cast on the polishing surface 2a. The area of ​​these shadows becomes smaller as the height of the convex portions 76 decreases. As a result, the intensity of the reflected light increases.

[0045] 10 is a schematic diagram showing an example of an image in which the brightness changes according to the change in the height of the convex portion 76. In the example shown in FIG. 10, the light projecting units 51 and 52 emit light at an incident angle of 30°. The low-brightness areas in each image in FIG. 10 correspond to the shadows of the convex portions 76 on the polishing surface 2a of the polishing pad 2. As shown in FIG. 10, the brightness of the image changes according to the change in the depth of the convex portions 76.

[0046] As described above, the light projecting units 51 and 52 simultaneously irradiate the target area T on the polishing surface 2a with multiple light beams of different wavelengths from different directions. In one embodiment, the light projecting units 51 and 52 emit light beams of different colors (i.e., different wavelengths). For example, the light projecting unit 51 emits one light beam from red, green, or blue, and the light projecting unit 52 emits another light beam from red, green, or blue. In the example described below, the light projecting unit 51 emits green light, and the light projecting unit 52 emits blue light. However, the colors of the light beams emitted by the light projecting units 51 and 52 are not limited to those in the embodiment described below.

[0047] 11 is a schematic diagram showing an example in which light is irradiated onto a target area T from two light-projecting units 51 and 52, two images are generated from reflected light corresponding to the light emitted from the two light-projecting units 51 and 52, and an evaluation image is created from these images. In the example shown in FIG. 11 , a recess 75 exists in the target area T.

[0048] The light-projecting units 51 and 52 simultaneously project green light and blue light onto the target area T. The imaging device 59 generates a first image from the reflected green light and a second image from the reflected blue light. The first image is an image generated from the reflected green light corresponding to the green light emitted from the light-projecting unit 51, and the second image is an image generated from the reflected blue light corresponding to the blue light emitted from the light-projecting unit 52.

[0049] 11 , the green light and the blue light are incident on the target area T from different directions, so the position of the shadow of the recess 75 in the first image is different from the position of the shadow of the recess 75 in the second image. The imaging device 59 simultaneously receives the green light and the blue light reflected from the target area T and simultaneously generates the first image and the second image from these reflected lights. Therefore, as shown in FIG. 11 , the recess 75 appears in the same position in the first image and the second image.

[0050] The system processor 70 generates an evaluation image, which is either a difference image or a division image, from the first image and the second image. A difference image is generated by calculating the absolute value of the difference between the luminance value of each pixel in the first image and the luminance value of the corresponding pixel in the second image. A division image is generated by dividing the luminance value of each pixel in the first image by the luminance value of the corresponding pixel in the second image. In the embodiment shown in Figure 11, a difference image is used as the evaluation image. In one embodiment, the evaluation image may be a division image.

[0051] As can be seen from Figure 11, a difference image is generated from two images generated simultaneously, thereby removing noise from the two images and canceling the brightness of areas other than the recessed portion 75. Therefore, the evaluation image formed from the difference image can represent minute changes in the brightness of the recessed portion 75, and the system processing unit 70 can compare a pad surface index value calculated based on the brightness information of the evaluation image with a threshold value to accurately detect changes in the properties of the polishing surface 2a of the polishing pad 2. Although not shown, a division image is also generated in a similar manner. In one embodiment, the system processing unit 70 can compare a pad surface index value calculated based on the brightness information of the evaluation image formed from the division image with a threshold value to accurately detect changes in the properties of the polishing surface 2a of the polishing pad 2.

[0052] To cancel the luminance of portions other than the recessed portion 75, the system processing unit 70 may change the density of the first image and the second image in advance. For example, the system processing unit 70 may add or subtract an offset value to the luminance values ​​of the first image and the second image, or normalize the luminance values ​​of the pixels. By adjusting the image density in this manner, it is possible to correct for differences in the overall intensity of reflected light depending on the irradiation direction and differences in the wavelength sensitivity of the image sensor 58.

[0053] The area of ​​the shadow of the recess 75 in the evaluation image decreases as the depth of the recess 75 decreases. Therefore, the brightness of the evaluation image decreases as the depth of the recess 75 decreases. The area of ​​the shadow is likely to change with the depth of the recess 75 when the angle of incidence θ of the light from the light projectors 51 and 52 onto the target area T is relatively small. In one embodiment, the angle of incidence θ of the light from the light projectors 51 and 52 onto the target area T is within a range of 3 to 80 degrees, preferably 3 to 45 degrees. The angle of incidence of the light from the light projector 51 onto the target area T is the same as the angle of incidence of the light from the light projector 52 onto the target area T. As long as the angle of incidence is within the above range, the angle of incidence of the light from the light projector 51 onto the target area T may be different from the angle of incidence of the light from the light projector 52 onto the target area T.

[0054] 12 is a schematic diagram showing another example in which light is irradiated onto a target area T from two light-projecting units 51, 52, two images are generated from reflected light corresponding to the light emitted from the two light-projecting units 51, 52, and an evaluation image is created from these images. In the example shown in FIG. 12, a convex portion 76 exists in the target area T. Operations not specifically described are the same as those in the embodiment described with reference to FIG. 11, and therefore redundant description will be omitted.

[0055] The light-projecting units 51 and 52 simultaneously project green light and blue light onto the target area T. The imaging device 59 generates a first image from the reflected green light and a second image from the reflected blue light. The first image is an image generated from the reflected green light corresponding to the green light emitted from the light-projecting unit 51, and the second image is an image generated from the reflected blue light corresponding to the blue light emitted from the light-projecting unit 52.

[0056] 12, the green light and the blue light are incident on the target area T from different directions, so the position of the shadow of the convex portion 76 in the first image is different from the position of the shadow of the convex portion 76 in the second image. The imaging device 59 simultaneously receives the green light and the blue light reflected from the target area T and simultaneously generates the first image and the second image from these reflected lights. Therefore, as shown in FIG. 12, the convex portion 76 appears in the same position in the first image and the second image.

[0057] The system processing unit 70 generates an evaluation image, which is either a difference image or a division image, from the first image and the second image. In the embodiment shown in FIG. 12 , a difference image is used as the evaluation image. In one embodiment, the evaluation image may be a division image. The area of ​​the shadow of the convex portion 76 in the evaluation image decreases as the height of the convex portion 76 decreases. Therefore, the brightness of the evaluation image decreases as the height of the convex portion 76 decreases. The area of ​​the shadow is likely to change with changes in the height of the convex portion 76 when the incident angle θ of the light from the light projectors 51 and 52 to the target area T is relatively small.

[0058] 13 is a schematic diagram showing an embodiment of a pad surface determination system 40 equipped with three light projecting units 51, 52, and 53. The configuration and operation of this embodiment not specifically described are the same as those of the embodiment described with reference to FIGS. 1 to 12, and therefore a duplicated description will be omitted.

[0059] The three light-projecting units 51, 52, and 53 are disposed above the target area T and face the target area T. The three light-projecting units 51, 52, and 53 are arranged at equal intervals around the center point of the target area T. In the example shown in FIG. 13 , a recess 75 is present within the target area T. The three light-projecting units 51, 52, and 53 irradiate the recess 75 within the target area T with light of different wavelengths (i.e., different colors), and the system processing unit 70 generates three images from reflected light corresponding to the light emitted from the three light-projecting units 51, 52, and 53, and creates an evaluation image from these images.

[0060] The three light-projecting units 51, 52, 53 simultaneously irradiate the target area T with first light, second light, and third light, and the imaging device 59 simultaneously receives the first reflected light, second reflected light, and third reflected light from the target area T and simultaneously generates a first image, a second image, and a third image from these reflected lights. Therefore, as shown in Fig. 13, the recess 75 appears at the same position in the first image, the second image, and the third image.

[0061] In one embodiment, the light projecting units 51, 52, and 53 simultaneously project red light, green light, and blue light onto the target area T. The light from the light projecting units 51, 52, and 53 overlaps within the target area T. As long as the light from the multiple light projecting units 51, 52, and 53 is simultaneously projected onto the target area T, the light emission start times and light emission stop times of the multiple light projecting units 51, 52, and 53 may be different.

[0062] The imaging device 59 generates a first image from the red reflected light, a second image from the green reflected light, and a third image from the blue reflected light. The first image is an image generated from the red reflected light corresponding to the red light emitted from the light-projecting unit 51, the second image is an image generated from the green reflected light corresponding to the green light emitted from the light-projecting unit 52, and the third image is an image generated from the blue reflected light corresponding to the blue light emitted from the light-projecting unit 53.

[0063] The system processing unit 70 generates an evaluation image, which is either a difference image or a division image, from the first, second, and third images. The difference image is created using the following formula: abs(GB)+abs(GR)+abs(RG), where abs represents the absolute value. The division image is created using the following formula: [(G / B)+(B / G)]+[(G / R)+(R / G)]+[(B / R)+(R / B)]. In these two formulas, R represents the luminance value of each pixel in the first image, G represents the luminance value of the corresponding pixel in the second image, and B represents the luminance value of the corresponding pixel in the third image.

[0064] In the embodiment shown in Figure 13, a difference image is used as the evaluation image. In one embodiment, the evaluation image may be a division image. The area of ​​the shadow of the recess 75 in the evaluation image decreases as the depth of the recess 75 decreases. Therefore, the brightness of the evaluation image decreases as the depth of the recess 75 decreases.

[0065] FIG. 14 is a schematic diagram showing an example in which light is irradiated from three light-projecting units 51, 52, and 53 onto a target area T where a convex portion 76 exists, three images are generated from reflected light corresponding to the light emitted from the three light-projecting units 51, 52, and 53, and an evaluation image is created from these images. The configuration and operation of this embodiment, unless otherwise specified, are the same as those of the embodiment described with reference to FIG. 13 . In the embodiment shown in FIG. 14 , a difference image is used as the evaluation image. In one embodiment, the evaluation image may be a division image. The area of ​​the shadow of the convex portion 76 in the evaluation image decreases as the height of the convex portion 76 decreases. Therefore, the brightness of the evaluation image decreases as the height of the convex portion 76 decreases.

[0066] In the embodiment described with reference to Figures 13 and 14, light of different wavelengths is irradiated onto the target area T from multiple directions, making it easier for the system processing unit 70 to detect small irregularities on the polishing surface 2a. In particular, the system processing unit 70 can detect minute scratches formed within the polishing surface 2a. In one embodiment, four or more light-projecting units emitting light of different wavelengths may be provided. An embodiment in which three or more light-projecting units emitting light of different wavelengths are provided can also be applied to the embodiments described below.

[0067] 11 to 14, and calculates a pad surface index value representing the properties of the polishing surface 2a based on the brightness information of the evaluation image. The evaluation image used to generate the pad surface index value may be the entire evaluation image or a part of it.

[0068] The pad surface index value is a numerical value calculated based on the luminance information of the evaluation image. The luminance information of the evaluation image may be luminance information of an evaluation image that has been shading-corrected to correct the luminance. The luminance information of the evaluation image may be luminance information of an evaluation image that has been cropped (trimmed) from the original evaluation image. Furthermore, image filtering (spatial filtering), such as smoothing to reduce noise, may be performed as preprocessing. In one example, shading correction may be performed on the evaluation image, and the shading-corrected evaluation image may be trimmed, or the shading-corrected evaluation image may be subjected to image filtering.

[0069] Examples of brightness information of the evaluation image include the brightness of the entire evaluation image, the brightness of a portion of the evaluation image, and the brightness distribution within the evaluation image. An example of the pad surface index value calculated based on the brightness information of the evaluation image may be the sum or statistical value (e.g., average, variance, standard deviation) of the brightness values ​​of the pixels that make up the evaluation image. In another example, the system processing unit 70 may perform image processing to identify recesses 75 or protrusions 76 of the polishing pad 2 in the evaluation image and calculate the pad surface index value based on the brightness within the recesses 75 or protrusions 76.

[0070] 15 , the pad surface index value may be a variance calculated from a plurality of brightness values ​​of a plurality of segment regions S in the evaluation image. The plurality of segment regions S may be, for example, grid-shaped regions predefined in the evaluation image. In another example, the plurality of segment regions S may be a plurality of regions of other shapes. The variance of the plurality of brightness values ​​of the plurality of segment regions S in the evaluation image is found as follows: First, a representative value such as the sum or statistical value (for example, average) of the plurality of brightness values ​​of the plurality of pixels in each segment region S is calculated, and then the variance of the plurality of representative values ​​calculated for each of the plurality of segment regions S is calculated. The variance calculated in this way is the pad surface index value of that evaluation image.

[0071] The variance calculated from the multiple brightness values ​​of the multiple segment regions S in the evaluation image represents the clarity of the recesses 75 or protrusions 76 that appear in the evaluation image. More specifically, when the recesses 75 or protrusions 76 appear clearly in the evaluation image (i.e., when the recesses 75 are deep or the protrusions 76 are high), the variance is large. In contrast, when the recesses 75 or protrusions 76 do not appear clearly in the evaluation image (i.e., when the recesses 75 are shallow or the protrusions 76 are low), the variance is small. Therefore, the system processing unit 70 may generate an alarm signal when the variance as the pad surface index value falls below a threshold. In one embodiment, the pad surface index value may be calculated based on the brightness values ​​of the evaluation image obtained by performing smoothing processing, rather than the above variance.

[0072] In yet another example, as shown in FIG. 16 , the pad surface index value may be the ratio of the luminance value of the first region S1 to the luminance value of the second region S2 in the evaluation image. The first region S1 is a region that includes a recess 75 or a protrusion 76. In the example shown in FIG. 16 , the first region S1 is located within the second region S2. In another example, the second region S2 may be located within the first region S1. In yet another example, the first region S1 and the second region S2 may partially overlap. In yet another example, the first region S1 and the second region S2 may be separated from each other. The second region S2 may not include the recess 75 or the protrusion 76. The ratio of the luminance value of the first region S1 to the luminance value of the second region S2 can be calculated by dividing the luminance value of the first region S1 by the luminance value of the second region S2. By performing such division, it is possible to cancel out changes over time in the light intensity of the light projecting units 51, 52 (and 53) and differences in the light intensity between the plurality of light projecting units 51, 52 (and 53).

[0073] Fig. 17 is a graph showing an example of a pad surface index value that changes with the usage time of the polishing pad 2. In Fig. 17, the vertical axis represents the pad surface index value, and the horizontal axis represents the usage time of the polishing pad 2. In the example shown in Fig. 17, the pad surface index value increases with the usage time of the polishing pad 2 and exceeds the threshold value. For example, as the usage time of the polishing pad 2 increases, wear of the polishing pad 2 and / or accumulation of polishing debris in the recesses 75 progresses (i.e., the depth of the recesses 75 decreases). In another example, as the usage time of the polishing pad 2 increases, the surface roughness of the polishing pad 2 decreases (i.e., the height of the protrusions 76 decreases).

[0074] When the pad surface index value changes beyond the threshold value, the system processing unit 70 can generate an alarm signal to notify that the life of the polishing pad 2 has expired. Depending on the algorithm for calculating the pad surface index value, the pad surface index value may decrease over time as the polishing pad 2 is used. In this case, the system processing unit 70 is configured to generate an alarm signal when the pad surface index value falls below the threshold value.

[0075] 11 to 14, the recessed portion 75 or the protruding portion 76 appears clearly on the evaluation image, but the recessed portion 75 or the protruding portion 76 may not appear clearly in the evaluation image if the exposure time of the image sensor 58 or the light emission time of the light-projecting portions 51, 52 (and 53) is too long compared to the rotation speed of the polishing table 3. Even in such cases, the evaluation image is generated from the light reflected from the target region T including the recessed portion 75 or the protruding portion 76, and therefore the overall brightness information of the evaluation image changes according to the change in the recessed portion 75 or the protruding portion 76.

[0076] The timing of irradiating the target area T with light and generating an image of the target area T is not particularly limited. However, in one embodiment, in order to accurately measure the intensity of the reflected light, the irradiation of the target area T with light and the generation of an image of the target area T are performed when a polishing liquid such as a slurry is removed from the polishing pad 2. For example, the irradiation of the target area T with light and the generation of an image of the target area T are performed during the dressing of the polishing pad 2, which is performed between the end of polishing of a substrate and the start of polishing of the next substrate. More specifically, the system processing unit 70 issues commands to the multiple light projecting units 51, 52 (and 53) to generate light during the dressing of the polishing pad 2, and the imaging device 59 receives multiple reflected lights from the target area T and generates multiple images from the multiple reflected lights corresponding to the multiple light projecting units 51, 52 (and 53). Since the light projecting units 51, 52 (and 53) emit light while the polishing pad 2 is rotating, the light is irradiated onto different areas on the polishing pad 2.

[0077] The shapes (grooves, holes, their widths and depths) of the recesses 75 formed on the polishing surface 2a of the polishing pad 2 differ depending on the type of polishing pad 2. According to this embodiment, the multiple light projecting units 51, 52 (and 53) irradiate the target area T with multiple beams of light from different directions, so that the pad surface index value exhibits a characteristic change according to the depth of the recesses 75 and the amount of polishing debris accumulated in the recesses 75, regardless of the type of polishing pad 2 and / or the position on the polishing surface 2a.

[0078] In one embodiment, the steps of irradiating the target area T with multiple lights from multiple light-projecting units 51, 52 (and 53), generating multiple images from the multiple reflected lights using an imaging device 59, generating an evaluation image from the multiple images, and calculating the pad surface index value based on the brightness information of the evaluation image are repeated multiple times while the polishing pad 2 is rotating, thereby calculating the pad surface index value from the multiple target areas T. This makes it possible to accurately evaluate the overall condition of the polishing surface 2a.

[0079] In the above-described embodiment, a single target area T is irradiated with multiple beams of light from different directions to calculate the pad surface index value. The beams irradiated from each direction cast shadows in different directions. The system processing unit 70 combines the shadows by generating a division image or a difference image from the multiple images in which the shadows appear. In other words, the shadow information reflecting the depth of the recesses 75 and the height of the protrusions 76 is amplified. Furthermore, the grooves of the polishing pad 2 are often formed with a certain directionality on the polishing surface 2a. However, by irradiating the light from multiple directions, shadow information according to each groove can be obtained. Even if the minute irregularities on the polishing surface 2a have a certain directionality, the pad surface index value can be calculated accurately in the same way.

[0080] In one embodiment, a plurality of light projecting units capable of irradiating light from a plurality of directions may be provided, and an image of the target area T may be generated by irradiating light from only one of the light projecting units, and the pad surface index value may be calculated from the image. In this case, the wavelength of light from the plurality of light projecting units may be the same. Since the pad surface index value is calculated from an image obtained when light is irradiated from a plurality of directions, it is possible to accurately calculate the pad surface index value even if the recesses 75 and protrusions 76 on the polishing pad 2 have directionality.

[0081] In one embodiment, a plurality of light projecting units capable of irradiating light from a plurality of directions may be provided, and an image of the target area T may be generated by irradiating light from only one of the light projecting units, followed by an image of the target area T by irradiating light from only the other light projecting units, and an evaluation image, which is a difference image or division image, may be created from the plurality of images. In this case, the wavelength of light from the plurality of light projecting units may be the same. It is desirable to generate an image of the target area T without rotating the polishing table 3 so that the target area T does not move.

[0082] In one embodiment, the following steps are repeated within a predetermined period: irradiating the target area T with multiple light beams from the multiple light-projecting units 51, 52 (and 53), generating multiple images from the multiple reflected light beams using the imaging device 59, generating an evaluation image from the multiple images, and calculating a pad surface index value based on brightness information of the evaluation image, thereby obtaining multiple pad surface index values. The system processing unit 70 calculates the variance of the multiple pad surface index values, and may generate an alarm signal when the variance is greater than a variance threshold. Examples of the predetermined period include the period required to polish a predetermined number of substrates and the period equivalent to one rotation of the polishing table 3.

[0083] As wear of the polishing pad 2 or accumulation of polishing debris in the recesses 75 of the polishing surface 2a progresses, local differences in wear and accumulation of polishing debris within the polishing surface 2a are expected to increase, resulting in an increase in the variance of the pad surface index values ​​obtained within the predetermined period. For example, there is a difference in the amount of wear of the polishing pad 2 between the region of the polishing pad 2 that contacts the center of the substrate and the region of the polishing pad 2 that contacts the edge of the substrate. Therefore, the variance of the multiple pad surface index values ​​obtained within the predetermined period is expected to increase. In another example, if the polishing pad 2 is excessively worn, the variance of the surface roughness of the polishing pad 2 is expected to increase within the polishing surface 2a. The system processing unit 70 generates an alarm signal when the variance of the multiple pad surface index values ​​is greater than the variance threshold, thereby notifying that the polishing pad 2 is significantly worn and / or that a large amount of polishing debris has accumulated within the recesses 75 of the polishing surface 2a.

[0084] In one embodiment, a plurality of pad surface determination systems 40 may be provided at different positions above the polishing surface 2a of the polishing pad 2. Any one of the plurality of pad surface determination systems 40 may acquire a plurality of pad surface index values ​​generated by the plurality of pad surface determination systems 40, calculate a variance of the plurality of pad surface index values, and generate an alarm signal when the variance is greater than a variance threshold. Furthermore, in one embodiment, the pad surface determination system 40 may move above the polishing pad 2 and emit light at different positions to generate a plurality of pad surface index values.

[0085] In the above-described embodiment, one imaging device 59 is provided, but multiple imaging devices 59 may be provided at different positions.

[0086] If there is foreign matter such as dust on the image sensor 58 of the imaging device 59, black spots M1 may appear in the image generated by the imaging device 59, as shown in Fig. 18. Furthermore, if there are air bubbles in the liquid on the polishing surface 2a of the polishing pad 2, white patterns M2 may appear in the image generated by the imaging device 59, as shown in Fig. 18. Such black spots M1 and white patterns M2 appear on the image as brightness deviation areas having brightness significantly different from other areas in the image. These brightness deviation areas prevent the pad surface determination system 40 from accurately determining the properties of the polishing surface 2a of the polishing pad 2.

[0087] Therefore, in one embodiment, the system processing unit 70 removes brightness deviation areas from the image as follows: The system processing unit 70 creates an evaluation image, which is either a difference image or a division image, from two images generated from the reflected light of the two light-projecting units 51, 52, and calculates the pad surface index value using brightness values ​​within an acceptable range of the brightness values ​​of the pixels constituting the evaluation image. The lower limit of the acceptable range is a threshold value for removing low brightness areas caused by foreign matter on the image sensor 58, and the upper limit of the acceptable range is a threshold value for removing high brightness areas caused by air bubbles in the liquid on the polishing surface 2 a.

[0088] In one embodiment, the allowable range is determined as follows. As shown in FIG. 19 , the system processing unit 70 creates a frequency distribution of the luminance values ​​of the pixels constituting the evaluation image, which is either a difference image or a division image, calculates a standard deviation σ from the frequency distribution, and determines the allowable range from the standard deviation σ. In one embodiment, the system processing unit 70 may determine the allowable range with a lower limit of −σ and an upper limit of +σ. In another embodiment, the system processing unit 70 may determine the allowable range with a lower limit of −3σ and an upper limit of +3σ. The center of the allowable range is, for example, the luminance value of peak P1 in the frequency distribution, or the median of the overall luminance values ​​of the evaluation image.

[0089] The system processing unit 70 calculates a pad surface index value using the brightness values ​​within the tolerance range determined as described above. Examples of the pad surface index value include the sum or statistical value (e.g., average, variance, standard deviation) of the brightness values ​​within the tolerance range. Since the brightness values ​​within the tolerance range do not include the brightness values ​​of pixels within the brightness deviation region, the pad surface determination system 40 can accurately determine the properties of the polishing surface 2 a of the polishing pad 2.

[0090] In another embodiment, the system processing unit 70 may remove brightness deviation regions from an image before generating an evaluation image, which is either a difference image or a division image, by following the same process as described above. For example, the system processing unit 70 corrects the first image generated from the reflected light of the light projector 51 by removing brightness values ​​outside an allowable range from the brightness values ​​of the pixels that make up the first image, and generates an evaluation image, which is either a difference image or a division image, from the corrected first image and the second image generated from the reflected light of the light projector 52.

[0091] The acceptable range can be determined in the same manner as in the embodiment described with reference to Fig. 19. That is, the system processing unit 70 creates a frequency distribution of the luminance values ​​of the pixels constituting the first image, calculates a standard deviation from the frequency distribution, and determines the acceptable range from the standard deviation. The center of the acceptable range is, for example, the peak luminance value in the frequency distribution of the first image or the median of the overall luminance values ​​of the first image.

[0092] The system processing unit 70 calculates a pad surface index value using the brightness values ​​of the pixels that make up the evaluation image generated from the corrected first image and second image. Examples of the pad surface index value include the sum or statistical value (e.g., mean, variance, standard deviation) of the brightness values ​​of the pixels that make up the evaluation image. Since the brightness values ​​of the pixels that make up the evaluation image do not include the brightness values ​​of pixels in the brightness deviation region, the pad surface evaluation system 40 can accurately evaluate the properties of the polishing surface 2 a of the polishing pad 2.

[0093] In one embodiment, the system processing unit 70 may correct the second image by removing brightness deviation areas from the second image before generating the evaluation image, similar to the first image. Specifically, the system processing unit 70 corrects the second image by removing brightness values ​​outside the allowable range from the brightness values ​​of pixels constituting the second image. The system processing unit 70 generates an evaluation image, which is either a difference image or a division image, from the corrected first image and the corrected second image, and calculates the pad surface index value using the brightness values ​​of the pixels constituting the evaluation image.

[0094] The tolerance range used for correcting the second image can be determined in the same manner as in the embodiment described with reference to Fig. 19. That is, the system processing unit 70 creates a frequency distribution of the luminance values ​​of the pixels constituting the second image, calculates a standard deviation from the frequency distribution, and determines the tolerance range from the standard deviation. The center of the tolerance range is, for example, the peak luminance value in the frequency distribution of the second image or the median of the overall luminance values ​​of the second image.

[0095] Next, still another embodiment of the pad surface determination system 40 will be described with reference to Figures 20 and 21. In this embodiment, the system processing unit 70 is configured to input the luminance information of the evaluation image to a determination model 78 constructed by machine learning, and to output a determination result of the surface texture of the polishing pad 2 from the determination model 78.

[0096] The system processing unit 70 has a determination model 78 stored in its storage device 70a. This determination model 78 is a trained model constructed by machine learning. Examples of machine learning include the support vector regression (SVR) method, the partial least squares (PLS) method, the deep learning method, the random forest method, and the decision tree method. In one example, the determination model 78 is composed of a neural network constructed by the deep learning method.

[0097] The training data used for machine learning of the determination model 78 includes evaluation images of the polishing surface of the polishing pad and also includes surface texture corresponding to the evaluation images of the polishing surface of the polishing pad, which is a correct label. The evaluation images included in the training data are generated in the same manner as in the above-described embodiment. That is, light is irradiated onto a target area within the polishing surface of the polishing pad from the two light-projecting units 51 and 52, two images are generated from the reflected light corresponding to the light emitted from the two light-projecting units 51 and 52, and an evaluation image, which is a difference image or a division image, is generated from these images.

[0098] The surface texture, which is the correct label, is a numerical value indicating the degree of the surface texture of the polishing pad, and can be expressed in a predetermined manner, such as 0 or 1, a percentage from 0 to 100%, a numerical value from 1 to 10, or a scale from 1 to 5. For example, an evaluation image generated when the worker determines that the polishing pad needs to be replaced is associated with a corresponding correct label of 0, and an evaluation image generated when the worker determines that the polishing pad does not need to be replaced is associated with a corresponding correct label of 1. Whether the polishing pad needs to be replaced is determined based on factors such as a decrease in the polishing rate and the amount of wear on the polishing pad.

[0099] In another example, when the surface texture as a correct label is expressed as a percentage of 0 to 100%, a surface texture of 0% indicates that the polishing pad is in a brand new state, and a surface texture of 100% indicates that the polishing pad needs to be replaced. An evaluation image generated when the operator determines that the polishing pad needs to be replaced is associated with a corresponding correct label of 100%. If the number of substrates polished when the operator determines that the polishing pad needs to be replaced is 1,000 (i.e., if the number of substrates polished is 1,000, corresponding to a correct label of 100%), an evaluation image of the polishing surface generated when 900 substrates have been polished is associated with a corresponding correct label of 90%. An evaluation image of the polishing surface generated when 800 substrates have been polished is associated with a corresponding correct label of 80%. Similarly, a correct label corresponding to an evaluation image of an intermediate state of the polishing surface of a polishing pad is determined. In this way, training data including correct labels of 0 to 100% and multiple evaluation images corresponding to these correct labels is obtained.

[0100] 21 is a schematic diagram showing an example of a determination model 78 constructed using a deep learning method. The determination model 78 has an input layer 101, multiple hidden layers (also called intermediate layers) 102, and an output layer 103. Luminance information of the evaluation image is input to the input layer 101, and a determination result of the surface texture of the polishing pad is output from the output layer 103. The determination result of the surface texture of the polishing pad output from the output layer 103 is, for example, a numerical value or a combination of multiple numerical values ​​indicating the surface texture of the polishing pad.

[0101] The determination model 78 is constructed using deep learning as follows. Luminance information of an evaluation image of the polishing surface of a polishing pad included in the training data is input to the input layer 101 shown in FIG. 21 . The determination model 78 is configured so that, when the luminance information of the evaluation image is input to the input layer 101, a numerical value indicating the surface texture of the polishing pad corresponding to the evaluation image is output from the output layer 103. In the machine learning for constructing the determination model 78, the system processing unit 70 compares the numerical value indicating the surface texture output from the output layer 103 with the correct label corresponding to the luminance information of the input evaluation image, and adjusts the parameters (weights, thresholds, etc.) of each node (neuron) to minimize the error. In this way, the determination model 78 is trained to output an appropriate surface texture determination result from the output layer 103 based on the luminance information of the evaluation image input to the input layer 101. The determination model 78 may be configured to output the determination result of the surface texture of the polishing pad along with the confidence level of the determination result.

[0102] According to the above embodiment, it is possible to monitor the surface texture of the polishing pad 2 based on the outputted surface texture determination result. For example, when the outputted surface texture is 50%, it can be determined that the current surface texture of the polishing pad 2 is half of the surface texture at the time of replacement of the polishing pad 2.

[0103] The above-described embodiments have been described for the purpose of enabling a person of ordinary skill in the art to practice the present invention. Various modifications of the above-described embodiments would be obvious to a person skilled in the art, and the technical concept of the present invention may be applied to other embodiments. Therefore, the present invention is not limited to the described embodiments, but is to be interpreted in the broadest scope in accordance with the technical concept defined by the claims.

[0104] The present invention can be used for a pad surface evaluation method and a pad surface evaluation system for evaluating the surface properties of a polishing pad used to polish a substrate such as a wafer.

[0105] REFERENCE SIGNS LIST 1 Polishing head 2 Polishing pad 2a Polishing surface 3 Polishing table 3a Table shaft 5 Polishing liquid supply nozzle 6 Table motor 10 Polishing head shaft 14 Support shaft 16 Polishing head swing arm 40 Pad surface determination system 50 Polishing control unit 51, 52, 53 Light projecting unit 58 Image sensor 59 Imaging device 70 System processing unit 75 Concave portion 76 Convex portion 78 Determination model T Target area

Claims

1. A pad surface evaluation method for evaluating the surface quality of a polishing pad having a polishing surface for polishing a substrate, comprising: simultaneously irradiating a target area within the polishing surface with multiple beams of light having different wavelengths from different directions when viewed from a direction perpendicular to the polishing surface; receiving multiple beams of light reflected from the target area with an imaging device; generating multiple images from the multiple beams of light corresponding to the multiple beams of light having different wavelengths with the imaging device; generating an evaluation image which is either a difference image or a division image of the multiple images; and evaluating the surface quality of the polishing pad based on the brightness information of the evaluation image.

2. The pad surface determination method according to claim 1, wherein the angles of incidence of the plurality of light beams with different wavelengths on the target area are within a range of 3 to 80 degrees.

3. The pad surface determination method according to claim 1, wherein the angles of incidence of the plurality of light beams with different wavelengths on the target area are the same.

4. The pad surface determination method according to claim 1, wherein the plurality of images are simultaneously generated by the imaging device.

5. A pad surface evaluation method as described in claim 1, wherein judging the surface quality of the polishing pad based on the brightness information of the evaluation image comprises calculating a pad surface index value representing the quality of the polishing surface based on the brightness information of the evaluation image.

6. A pad surface judgment method according to claim 5, further comprising: acquiring a plurality of pad surface index values ​​by repeating the steps of irradiating the target area with a plurality of light beams having different wavelengths, generating a plurality of images from the plurality of reflected light beams using the imaging device, generating the evaluation image from the plurality of images, and calculating the pad surface index value based on brightness information of the evaluation image; and calculating the variance of the plurality of pad surface index values.

7. A pad surface determination method as described in claim 5, wherein the step of calculating the pad surface index value based on the brightness information of the evaluation image is a step of calculating the pad surface index value using brightness values ​​within an allowable range among the brightness values ​​of the pixels constituting the evaluation image.

8. The pad surface evaluation method according to claim 7, further comprising: creating a frequency distribution of the brightness values ​​of the pixels that make up the evaluation image; calculating a standard deviation from the frequency distribution; and determining the tolerance range from the standard deviation.

9. A pad surface evaluation method as described in claim 1, wherein the step of generating the evaluation image from the plurality of images is a step of correcting at least one of the plurality of images by removing brightness values ​​outside an allowable range from brightness values ​​of pixels constituting at least one of the plurality of images, and generating the evaluation image, which is either a difference image or a division image, from the plurality of images including the corrected image.

10. The pad surface evaluation method according to claim 9, further comprising: creating a frequency distribution of the brightness values ​​of the pixels constituting said at least one image; calculating a standard deviation from said frequency distribution; and determining said tolerance range from said standard deviation.

11. The pad surface evaluation method according to claim 1, wherein judging the surface quality of the polishing pad based on the brightness information of the evaluation image comprises inputting the brightness information of the evaluation image into a judgment model constructed by machine learning, and outputting a judgment result of the surface quality of the polishing pad from the judgment model.

12. A pad surface evaluation system for evaluating the surface quality of a polishing pad having a polishing surface for polishing a substrate, comprising: a plurality of light projecting units that simultaneously irradiate a target area within the polishing surface with a plurality of light beams of different wavelengths from different directions when viewed from a direction perpendicular to the polishing surface; an imaging device that receives a plurality of reflected light beams from the target area and generates a plurality of images from the reflected light beams corresponding to the plurality of light beams of different wavelengths; and a system processing unit that generates an evaluation image, which is either a difference image or a division image, from the plurality of images and evaluates the surface quality of the polishing pad based on the brightness information of the evaluation image.

13. The pad surface determination system according to claim 12, wherein the angles of incidence of the plurality of light beams with different wavelengths on the target area are within a range of 3 to 80 degrees.

14. The pad surface determination system according to claim 12, wherein the angles of incidence of the plurality of lights with different wavelengths on the target area are the same.

15. The pad surface determination system according to claim 12, wherein the imaging device is configured to generate the plurality of images simultaneously.

16. The pad surface determination system according to claim 12, wherein the system processing unit is configured to calculate a pad surface index value representing the properties of the polishing surface based on brightness information of the evaluation image.

17. The pad surface determination system according to claim 16, wherein the system processing unit is configured to calculate the variance of a plurality of pad surface index values ​​calculated at different times.

18. A pad surface determination system as described in claim 16, wherein the system processing unit is configured to calculate the pad surface index value using brightness values ​​within an allowable range of brightness values ​​of pixels constituting the evaluation image.

19. The pad surface evaluation system according to claim 18, wherein the system processing unit is configured to: create a frequency distribution of the brightness values ​​of the pixels that make up the evaluation image; calculate a standard deviation from the frequency distribution; and determine the tolerance range from the standard deviation.

20. The pad surface determination system of claim 12, wherein the system processing unit is configured to: correct at least one of the plurality of images by removing brightness values ​​outside an acceptable range from the brightness values ​​of pixels constituting at least one of the plurality of images; and generate the evaluation image, which is either a difference image or a division image, from the plurality of images including the corrected image.

21. The pad surface determination system according to claim 20, wherein the system processing unit is configured to: create a frequency distribution of brightness values ​​of pixels constituting the at least one image; calculate a standard deviation from the frequency distribution; and determine the tolerance range from the standard deviation.

22. The pad surface judgment system described in claim 12, wherein the system processing unit is configured to input the brightness information of the evaluation image into a judgment model constructed by machine learning, and output the judgment result of the surface property of the polishing pad from the judgment model.

Images