Monitoring of Polishing Pad Texture in Chemical Mechanical Polishing

Abstract

To provide an apparatus that can avoid contamination and damage to a polishing pad, and can reduce wafer-to-wafer non-uniformity (WTWNU).SOLUTION: An apparatus 20 for chemical mechanical polishing includes a platen 24 having a surface to support a polishing pad 30, a carrier head 70 to hold a substrate 10 against a polishing surface 36 of the polishing pad 30, a pad conditioner 60 to press abrasive bodies against the polishing surface 36, an in-situ polishing pad monitoring system 40 including an imager 42 disposed above the platen 24 to capture an image of the polishing pad 30, and a controller 90 configured to receive the image from the monitoring system 40 and generate a measure of polishing pad surface roughness based on the image. The controller 90 can use machine learning-based image processing to generate the measure of surface roughness.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to optical monitoring of polishing pads used in chemical mechanical polishing.

Background Art

[0002] Integrated circuits are typically formed on a substrate by continuously depositing conductive, semiconductive, or insulating layers on a silicon wafer. In a variety of manufacturing processes, planarization of the layers on the substrate is required. For example, one manufacturing step involves depositing a conductive fill layer on a patterned insulating layer to fill trenches or holes in the insulating layer. The fill layer is then polished until the convex pattern of the insulating layer is exposed. After planarization, the portions of the conductive fill layer remaining between the convex patterns of the insulating layer form vias, plugs, and lines that provide conductive paths between thin film circuits on the substrate.

[0003] Chemical mechanical polishing (CMP) is a generally recognized planarization method. In this planarization method, typically, the substrate needs to be mounted on a carrier head. The exposed surface of the substrate is arranged to abut against a rotating polishing pad. The carrier head applies a controllable load to the substrate to press the substrate against the polishing pad. A polishing liquid (such as a slurry having abrasive grains) is supplied to the surface of the polishing pad.

[0004] After the CMP process is carried out for a certain period of time, the surface of the polishing pad may become glazed (clogged) due to the accumulation of the removed material and / or slurry by-products from the substrate and / or the polishing pad. Glazing may cause the polishing rate to slow down or the non-uniformity on the substrate to increase.

[0005] Typically, the polishing pad is maintained at a desired surface roughness (and glazing is avoided) by a conditioning process using a pad conditioner. The pad conditioner is used to remove unwanted deposits on the polishing pad and to regenerate the surface of the polishing pad to a desired asperity. A typical pad conditioner generally includes an abrasive head embedded with diamond abrasive grains, which can be abutted against and rubbed on the polishing pad surface to restore the texture of the pad. SUMMARY OF THE INVENTION

[0006] In one aspect, an apparatus for chemical mechanical polishing includes a platen having a surface for supporting a polishing pad, a carrier head for holding a substrate in contact with the polishing surface of the polishing pad, a pad conditioner for pressing abrasive grains in contact with the polishing surface, an in-situ polishing pad monitoring system including an imaging device disposed above the platen for capturing an image of the polishing pad, and a controller configured to receive the image from the monitoring system and generate a measure of the polishing pad surface roughness based on the image.

[0007] The implementation may include one or more of the following features.

[0008] The controller may operate as a machine learning-based image processing system and may be configured to input the image to the image processing system. The machine learning-based image processing system may include a supervised learning module. The machine learning-based image processing system may include a dimensional reduction module for receiving the image and an output component value, and the controller may be configured to input the component value for the image to the supervised learning module. The controller may be configured to directly input the image to the supervised learning module. The controller may be configured to operate the supervised learning module as an artificial neural network.

[0009] The controller may be configured to receive other data including values for parameters and may be configured to generate a reference for the polishing pad surface roughness based on the image and the parameter values. The parameters may be polishing control parameters, state parameters, measurements from sensors within the polishing system, or measurements of the polishing pad by sensors outside the polishing system. The parameters may be the platen rotation speed, the slurry dispensing speed, the slurry composition, the number of substrates since the polishing pad was changed, or the measured surface roughness of the polishing pad by a stand-alone measurement station before the polishing pad is mounted on the platen.

[0010] The controller may be configured to at least one of stop the conditioning process or adjust the conditioning parameters based on the reference for the polishing pad surface roughness.

[0011] The imaging device may be movable radially on the platen. The imaging device can be mounted on an arm that can swing laterally on the platen.

[0012] In another aspect, a polishing method includes contacting a substrate with a polishing pad on a platen, generating relative movement between the substrate and the polishing pad, capturing an image of the polishing pad with an optical sensor, and generating a measured value of the surface roughness of the polishing pad by inputting the image into a machine learning-based image processing system.

[0013] Embodiments may include one or more of the following features.

[0014] It is possible to receive training data including pairs of a plurality of training images and training values. The supervised learning algorithm in the learning-based image processing system may be trained using the training data. The conditioning process may be stopped, or the conditioning parameters may be adjusted based on the reference for the polishing pad surface roughness.

[0015] Certain implementations include one or more of the following advantages, but are not limited thereto. Since the roughness of the polishing pad can be determined using non-contact techniques, contamination and damage to the polishing pad can be avoided. The roughness of the polishing pad can be determined accurately and quickly, and the conditioning process can be adjusted appropriately. Wafer-to-wafer non-uniformity (WTWNU) can be reduced. Since the roughness can be determined using non-contact techniques, contamination of the polishing pad can be avoided.

[0016] Details of one or more implementations are set forth in the accompanying drawings and the following description. Other aspects, features, and advantages will be apparent from this description and the drawings, as well as from the claims.

Brief Description of the Drawings

[0017]

Figure 1

Figure 2

Figure 3

Modes for Carrying Out the Invention

[0018] Similar reference symbols in the various drawings indicate similar elements.

[0019] The chemical mechanical polishing process tends to reduce the surface roughness of the polishing pad, for example, due to the glazing effect described above. Conditioning can be used to restore the surface roughness. However, the extent to which the pad is glazed, as well as The degree to which conditioning restores surface roughness can be non-uniform across the entire polishing pad. As a result, even after conditioning, non-uniformities can occur in the surface roughness of the polishing pad. Further, conditioning techniques can wear the pad at different rates across the pad, creating non-uniformities in the pad thickness and leaving periodic scratching or scoring on the polishing pad surface. To measure surface roughness, contact techniques, such as profilometers, can be used, but this can pose a risk of contamination and may not be practical at all for application to pads already provided with the required equipment (conditioners, carriers, etc.). However, the polishing pad can be imaged, and the image can be sent to a trained machine learning model that outputs measurements of the surface texture. The controller can then use these measurements to adjust the conditioning process to achieve a target surface texture or to improve the uniformity of the surface texture across the entire polishing pad.

[0020] The term "surface texture" is used herein to include surface roughness such as Ra, Rms, RSk, or Rp, and other irregularities of the polishing pad surface such as waviness smaller than the normal groove or perforation pattern on the polishing pad. For example, assuming a groove depth of 20 mils, the surface texture can include irregularities up to about 40 - 50 microns.

[0021] FIGS. 1 and 2 illustrate an example of a polishing system 20 of a chemical mechanical polishing apparatus. The polishing system 20 includes a rotatable disk-shaped platen 24 with a polishing pad 30 placed thereon. The platen 24 is operable to rotate about an axis 25. To rotate the platen 24, for example, a motor 22 can rotate a drive shaft 28. The polishing pad 30 can be a two-layer polishing pad having an outer layer 34 and a more flexible backing layer 32. The upper surface of the polishing pad 30 provides a polishing surface 36.

[0022] The polishing system 20 may include a supply port or an integrated supply - cleaning arm 39 for dispensing a polishing liquid 38 (such as slurry) onto the polishing pad 30.

[0023] The polishing system 20 may also include a polishing pad conditioner 60 for truing the polishing pad 30 and maintaining the polishing surface 36 in a consistent abrasive state. The polishing pad conditioner 60 includes a base, an arm 62 that can sweep laterally across the polishing pad 30, and a conditioner head 64 connected to the base by the arm 62. The conditioner head 64 conditions the polishing pad 30 by bringing an abrasive surface (for example, the lower surface of a disk 66 held by the conditioner head 64) into contact with the polishing pad 30. The abrasive surface may be rotatable, and the pressure of the abrasive surface against the polishing pad may be controllable.

[0024] In some implementations, the arm 62 is pivotally attached to the base and sweeps back and forth to move the conditioner head 64 in an oscillatory sweep motion across the polishing pad 30. The motion of the conditioner head 64 may be synchronized with the motion of the carrier head 70 to prevent collisions.

[0025] The vertical motion of the conditioner head 64 and the control of the pressure of the conditioning surface against the polishing pad 30 may be provided by a vertical actuator 68 above or within the conditioner head 64 (for example, a pressurizable chamber arranged to apply a downward pressure to the conditioner head 64). Alternatively, the vertical motion and pressure control may be provided by a vertical actuator within the base that raises the entire arm 62 and conditioner head 64, or by a pivotal connection between the arm 62 and the base that allows the tilt angle of the arm 62, and thus the height of the conditioner head 64 above the polishing pad 30, to be controllable.

[0026] The carrier head 70 is operable to hold the substrate 10 in contact with the polishing pad 30. The carrier head 70 is suspended from a support structure 72 (such as a carousel or track), and is connected by a drive shaft 74 to a carrier head rotation motor 76 so that the carrier head can rotate about an axis 71. Optionally, the carrier head 70 can oscillate laterally (e.g., along a slider of the carousel or track 72), or by the rotational rocking of the carousel itself. During operation, the platen rotates about its central axis 25, the carrier head rotates about its central axis 71, and translates horizontally across the upper surface of the polishing pad 30. The carrier head 70 can include a flexible membrane 80 having a substrate mounting surface that contacts the back surface of the substrate 10, and a plurality of pressurizable chambers 82 for applying different pressures to various zones (e.g., various radial zones) on the substrate 10. The carrier head can also include a retaining ring 84 for holding the substrate.

[0027] The polishing system 20 includes an in-situ optical pad monitoring system 40 that generates a signal representative of the surface texture (e.g., surface roughness) of the polishing pad 30. The in-situ optical pad monitoring system 40 includes an imaging device 42 (such as a camera disposed above the polishing pad, e.g., on a support arm 44). For example, the imaging device 42 can be a line scan camera, and the pad monitoring system 40 can be configured to generate a 2D image from a plurality of measurements by the line scan camera as the polishing pad 30 sweeps under the camera 40 due to the rotation of the platen 24. Alternatively, the imaging device 42 can be a 2D camera. The imaging device 42 can have a field of view 43 of a portion of the surface 36 of the polishing pad 30. The camera can include a CCD array and optical components (e.g., lenses) for focusing the imaging surface on the surface 36 of the polishing pad 30.

[0028] In some implementations, the imaging device 42 is disposed at a fixed radial position and images a fixed radial zone of the polishing pad 30. In this situation, the in-situ pad monitoring system 40 can generate measurements of the surface texture (e.g., surface roughness) at a fixed radial position on the polishing pad 30.

[0029] However, in some implementations, the imaging device 42 is movable, for example, laterally along the radius of the polishing pad 30. For example, referring to FIG. 2, the base 46 that holds the support arm 42 is configured to pivot, thereby causing the arm 42 (see arrow A) to swing across the polishing pad 30 and carry the imaging device 42 along an arcuate path. As another example, the support arm 44 may be a linear rail or may include a linear rail, and the imaging device 42 may be movable along the rail by a linear actuator 46 such as a stepper motor having a linear screw. By taking images of the polishing pad 30 at different radial zones, the in-situ pad monitoring system 40 can generate measurements of the surface texture (e.g., surface roughness) at different radial positions on the polishing pad 30.

[0030] The controller 90 (e.g., a general-purpose programmable digital computer) can receive images from the in-situ polishing pad monitoring system 40 and be configured to generate a reference for the surface texture (e.g., surface roughness) of the polishing pad 30 from this image. In this regard, the controller 90 (or the portion of the software that provides the surface texture measurements) can be considered part of the pad monitoring system 40. As described above, the surface roughness of the polishing pad 30 changes over time (e.g., during the process of polishing a plurality of substrates) due to the polishing and conditioning processes.

[0031] In addition, the controller 90 can be configured to control the pad conditioner 60 system based on the value of the surface texture (e.g., surface roughness) received from the in-situ pad monitoring system 40. For example, when the standard of the surface texture of the polishing pad 30 meets a threshold value, the controller 90 can stop the conditioning process. As another example, when the surface texture of the polishing pad meets another threshold value, the controller 90 can issue a warning to the operator of the polishing system 20 that, for example, the polishing or conditioning operation is not proceeding as expected.

[0032] As another example, when the in-situ pad monitoring system 40 generates measurement values regarding the surface texture (e.g., surface roughness) at different radial positions (with respect to the rotation axis 25) on the polishing pad 30, the controller 90 can use those measurement values to control the pad conditioner 60 and improve the uniformity of the surface texture (e.g., surface roughness). For example, the controller 90 can control the sweep of the conditioner arm 62 to control the residence time of the conditioner disk 64 in different radial zones on the polishing pad. For example, when it is necessary to increase the surface roughness within a radial zone, the residence time can be lengthened, while when it is necessary to decrease the surface roughness within a radial zone, the residence time can be shortened.

[0033] Referring to FIG. 3, an image from the in-situ pad monitoring system 40 is supplied to the trained machine vision image processing system 100. The machine vision image processing system 100 is configured to output a value representing the texture (e.g., surface roughness) of a portion of the polishing surface 36 within the field of view 43 of the imaging device 42. The machine vision image processing system 100 can be implemented as part of the controller 90. The machine vision image processing system 100 can incorporate various machine learning techniques. For example, the machine vision image processing system 100 can include a neural network, although other approaches, such as a naive Bayes classifier or a support vector machine, are also possible.

[0034] FIG. 3 shows functional blocks that can be implemented for the machine learning-based image processing system 100. These functional blocks can include an optional dimensionality reduction module 110 for performing dimensionality reduction of the image, and a supervised learning module 120 (which is shown implemented as a neural network). The supervised learning module 120 implements a supervised learning algorithm to generate a function that outputs a measurement of the surface texture (e.g., surface roughness) based on the image (or the data with dimensionality reduced from the image). As described above, these functional blocks can be distributed across multiple computers.

[0035] The output of the supervised learning module 120 can be supplied to a process control system 130 that can be implemented as part of the controller 90 to adjust the polishing process based on the measurement of the surface texture. For example, the process control system 130 can detect a conditioning end point and stop conditioning during the polishing process, and / or adjust conditioning parameters (e.g., sweep profile, conditioner head pressure, etc.) to reduce the non-uniformity of the surface texture (e.g., surface roughness) of the polishing surface 36 based on the measurement of the surface texture (e.g., surface roughness).

[0036] Assuming that the machine learning module 120 is a neural network, the neural network includes a plurality of input nodes 122 for each of the principal components, a plurality of hidden nodes 124 (hereinafter also referred to as "intermediate nodes"), and one output node 126 that generates a measurement of the surface texture (e.g., surface roughness). Generally, the hidden node 124 outputs a value that is a non-linear function of the weighted sum of the values from the input nodes 122 to which the hidden node is connected.

[0037] For example, the output of a certain hidden node 124 (designated by node k) can be expressed as follows. tanh(0.5*a k1 (I1)+a k2 (I2)+…+a kM (I M )+b k ) Equation 1 Here, tanh is the hyperbolic tangent, a kx is the weight of the connection between the k-th intermediate node and the x-th input node (among the M input nodes), and I M is the value of the M-th input node. However, other non-linear functions (e.g., the rectified linear unit (ReLU) function and its variants) can also be used instead of tanh.

[0038] The optional dimensionality reduction module 110 reduces the image so that the component values 112 (e.g., L component values) become a more limited number. The neural network 120 includes input nodes 122 for each component for which the image is reduced. For example, when the dimensionality reduction module 110 generates L component values, the neural network 120 includes at least input nodes N1, N2…N L and so on.

[0039] However, the supervised learning module 120 can optionally receive one or more inputs 114 other than the image or component values. The other inputs 114 can include measurements from other sensors within the polishing system, such as the temperature of the pad measured by a temperature sensor, or the slurry flow rate measured from a flow sensor. The other inputs can include values of polishing control parameters, such as platen rotation speed, slurry flow rate, or slurry composition. The polishing control parameter values can be obtained from a polishing recipe stored by the controller 90. The other inputs can include state parameters tracked by the controller, such as the identification of the various pads being used (manufacturer, brand name, pad composition, groove pattern, etc.), or the number of substrates polished since the pad was changed. The other inputs can include measurements from sensors that are not part of the polishing system, such as the measurement of the surface texture (e.g., surface roughness) of the polishing pad by a stand-alone measurement station before the pad is mounted on the platen. Thereby, the supervised learning module 120 can consider these other processing or environmental variables in the calculation of the surface texture (e.g., surface roughness). Assuming that the supervised learning module 120 is a neural network, the neural network can include one or more other input nodes (e.g., node 122a) for receiving the other data.

[0040] The architecture of the neural network 120 can vary in depth and width. For example, although the neural network 120 is illustrated with a single row of hidden nodes 124, it may include multiple rows. The number of intermediate nodes 124 can be equal to or greater than the number of input nodes 122. The neural network can be fully connected or a convolutional network.

[0041] For example, it is necessary to configure the supervised learning module 120 before it is used for the processing of device wafers.

[0042] As part of the construction procedure, the supervised learning module 120 receives training data that may include a plurality of training images and a plurality of training values, such as surface texture values (such as surface roughness values). Each reference image has a training value. That is, the training data includes pairs of images and training values.

[0043] For example, the images can be obtained from various pad samples. In addition, the measurement of the surface roughness of the sample can be performed with a measuring device, such as a contact profilometer, an interferometer, or a confocal microscope. Therefore, each training image can be associated with a training value that is the surface roughness of the sample from which the image was taken.

[0044] In some implementations, the data storage device can store multiple sets of training data. Different sets of training data can correspond to different types of polishing pads, such as different compositions and / or groove patterns. The supervised learning module 120 can receive various sets of training data from the operator of the semiconductor manufacturing plant, for example, via a user interface.

[0045] The training of the supervised learning module 120 can be performed using conventional techniques. For example, in the case of a neural network, the training can be performed by backpropagation using the training images and training values. For example, while the neural network is operating in the training mode, the reduced dimensional values of the training images are supplied to the respective input nodes N1, N2…N L and the training value V is supplied to the output node 126. This can be repeated for each pair of image and training value. If the supervised learning module 120 receives inputs other than images or component values, the values of these parameters can also be supplied to the machine learning module as training data.

[0046] When the training is executed, next, for example, as described above, the trained instantiation of the supervised learning module can be used. That is, during the processing of the substrate, an image of the polishing pad as well as other parameter values can be sent to the trained supervised learning module 120, and this learning module outputs a value for the surface texture (e.g., surface roughness). Then, the value of the surface texture (e.g., surface roughness) can be used, for example, as described above, to control the conditioning operation. The in-situ polishing pad monitoring system can be used in various polishing systems. To provide relative movement between the polishing surface and the substrate, either or both of the polishing pad and the carrier head can move. The polishing pad can be a circular (or some other shape) pad fixed to the platen, a tape extending between a supply roller and a take-up roller, or a continuous belt. The polishing pad can be fixed on the platen, progressive on the platen during the polishing operation and operation, or continuously driven on the platen during polishing. The pad can be fixed to the platen during polishing, or there can be a fluid bearing between the platen and the polishing pad during polishing. The polishing pad can be a standard (e.g., polyurethane with or without filler) rough pad, a soft pad, or a fixed abrasive pad.

[0047] In addition, although the foregoing description focuses on monitoring during polishing, it is also possible to obtain measurement values of the polishing pad before or after the substrate is polished (e.g., while the substrate is being transferred to the polishing system).

[0048] The controller and its functional operations can be implemented in digital electronic circuitry, or in computer software, firmware, or hardware, or in combinations thereof. A controller “configured” to perform the operations has software, firmware, or hardware sufficient to actually perform the operations and can be merely programmed or reprogrammed to perform the operations.

[0049] Embodiments can be implemented as one or more computer program products, i.e., one or more computer programs tangibly embodied in an information carrier (e.g., a non-transitory machine-readable storage medium, or a propagated signal) for execution by, or to control the operation of, a data processing apparatus (e.g., a programmable processor, a computer, or multiple processors or computers). The computer programs (also known as programs, software, software applications, or code) can be written in any form of programming language, including compiled or interpreted languages, and can be deployed as a stand-alone program or in any other form suitable for use in a module, component, subroutine, or other unit in a computing environment. One computer program need not correspond to one file. The program can be stored in a portion of a file that holds other programs or data, in a single file dedicated to the program being targeted, or in multiple coordinated files (e.g., files that store one or more modules, sub-programs, or portions of code). The computer program can be deployed to be executed on one computer or on multiple computers interconnected by a communication network and located at one site or distributed across multiple sites.

[0050] The processes and logical flows described in this specification may be implemented by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logical flows may also be implemented by special purpose logic circuitry, such as an FPGA (Field Programmable Gate Array) or ASIC (Application Specific Integrated Circuit), and the apparatus may be implemented as such special purpose logic circuitry.

[0051] Some embodiments of the invention have been described. It will be understood that various modifications may be made without departing from the spirit and scope of the invention. Accordingly, other embodiments are within the scope of the following claims.

Claims

1. 1. An apparatus for chemical mechanical polishing, comprising: a platen having a surface for supporting a polishing pad; a carrier head for holding a substrate in contact with the polishing surface of the polishing pad; a pad conditioner for contacting the polishing surface and pressing the abrasive grains; an in situ polishing pad monitoring system including an imaging device positioned above the platen to capture an image of the polishing pad; a controller configured to receive the images and other data from the monitoring system, including values ​​of parameters, and generate a measurement of a polishing pad surface texture based on the images and the values ​​of the parameters; Including, The parameters include platen rotation speed, slurry dispense rate, slurry composition, number of substrates since the polishing pad was changed, or a measurement of the surface roughness of the polishing pad by a stand-alone metrology station before the polishing pad was mounted on the platen.

2. The apparatus of claim 1 , wherein the parameters further include polishing control parameters, state parameters, measurements from a sensor within the polishing system, or measurements of the polishing pad by a sensor external to the polishing system.

3. 10. The apparatus of claim 1, wherein the controller is configured to at least one of: stop a conditioning process or adjust conditioning parameters based on the measurement of polishing pad surface texture.

4. The apparatus of claim 1 , wherein the controller is configured to operate as a machine learning based image processing system and to input the image into the image processing system.

5. The apparatus of claim 4 , wherein the machine learning based image processing system includes a supervised learning module.

6. 6. The apparatus of claim 5, wherein the machine learning based image processing system comprises a dimensionality reduction module that receives the image and output component values, and the controller is configured to input the component values ​​for the image to the supervised learning module.

7. The apparatus of claim 5 , wherein the controller is configured to input the images directly into the supervised learning module.

8. The apparatus of claim 5 , wherein the controller is configured to operate the supervised learning module as an artificial neural network.

9. 1. An apparatus for chemical mechanical polishing, comprising: a platen having a surface for supporting a polishing pad; a carrier head for holding a substrate in contact with the polishing surface of the polishing pad; a pad conditioner for contacting the polishing surface and pressing the abrasive grains; an in situ polishing pad monitoring system including an imaging device disposed above the platen and movable radially above the platen to capture an image of the polishing pad; a controller configured to receive the images and other data from the monitoring system, the data including values ​​of parameters; Including, The parameters include platen rotation speed, slurry dispense rate, slurry composition, number of substrates since the polishing pad was changed, or a measurement of the surface roughness of the polishing pad by a stand-alone metrology station before the polishing pad was mounted on the platen.

10. The apparatus of claim 9 , wherein the imaging device is mounted on a swingable arm that can swing laterally above the platen.

11. contacting the substrate with a polishing pad on a platen; generating relative motion between the substrate and the polishing pad; capturing an image of the polishing pad with an optical sensor; receiving other data including values ​​of the parameters; generating a measurement of a surface texture of the polishing pad by inputting the image and the parameter values ​​into a machine learning based image processing system; Including, The method of polishing, wherein the parameters include platen rotation speed, slurry dispensing rate, slurry composition, number of substrates since the polishing pad was changed, or a measurement of the surface roughness of the polishing pad by a stand-alone metrology station before the polishing pad was mounted on the platen.

12. The method of claim 11 , wherein the parameters further include polishing control parameters, state parameters, measurements from a sensor within the polishing system, or measurements of the polishing pad by a sensor outside the polishing system.

13. 12. The method of claim 11, comprising receiving training data comprising a plurality of pairs of training images and training values, and using the training data to train a supervised learning algorithm in the machine learning based image processing system.

14. The method of claim 13 , wherein the training values ​​include values ​​of a surface texture.

15. The method of claim 14 , wherein the training values ​​include surface roughness values.

16. 12. The method of claim 11, comprising stopping the conditioning process or adjusting conditioning parameters based on the measurement of the polishing pad surface texture.

Images