A polishing apparatus includes: a polishing table 12 for holding a polishing pad; a first electric motor 14 that rotationally drives the polishing table 12; a first rotary joint 40 that has a rotating body 41 that is rotationally driven by the first electric motor 14, a housing 42 provided around the rotating body 41, and a seal portion 44 that seals between the rotating body 41 and the housing 42; a second current sensor 31 that detects a current which is correlated with driving load of the first electric motor 14; a friction detecting unit 50 that detects sliding friction in the seal portion 44 of the first rotary joint 40; and an end point detecting apparatus 60 that detects a polishing end point of the polishing target on the basis of the current and the sliding friction.

It is possible to save labor, energy, and/or cost for a substrate processing apparatus.

It is provided with a sensor installed in a substrate processing apparatus and configured to detect a target physical quantity during processing of a target substrate; and a prediction unit configured to output a polishing end point timing, which is timing of ending polishing, by inputting, to a learned machine learning model, time-series data of the physical quantity detected by the sensor or time-series data obtained by differentiating the time-series data of the physical quantity with respect to time, in which the machine learning model is obtained by machine learning using, as a learning data set, past time-series data of the physical quantity or time-series data obtained by differentiating the past time-series data of the physical quantity with respect to time as input and using the past polishing end point timing as output.

It is possible to save labor, energy, and/or cost for a substrate processing apparatus.

It is provided with a sensor installed in a substrate processing apparatus and configured to detect a target physical quantity during processing of a target substrate; and a prediction unit configured to output a polishing end point timing, which is timing of ending polishing, by inputting, to a learned machine learning model, time-series data of the physical quantity detected by the sensor or time-series data obtained by differentiating the time-series data of the physical quantity with respect to time, in which the machine learning model is obtained by machine learning using, as a learning data set, past time-series data of the physical quantity or time-series data obtained by differentiating the past time-series data of the physical quantity with respect to time as input and using the past polishing end point timing as output.

A carrier head for holding a substrate in a polishing system includes a housing, a first flexible membrane secured to the housing to form one or more pressurizable chambers to apply pressure through a central membrane portion of the first flexible membrane to a central portion of a substrate, and a plurality of independently operable piezoelectric actuators supported by the housing, the plurality of piezoelectric actuators positioned radially outward of the central membrane portion and at different angular positions so as to independently adjust pressure on a plurality of angular zones in an annular outer region of the substrate surrounding the central portion of the substrate.

A polishing method capable of measuring a film thickness of a substrate, such as a semiconductor wafer, having various structural elements on its surface with high accuracy is disclosed. The polishing method includes: generating spectra of reflected lights from measurement points on a substrate; classifying the spectra based on a shape of each spectrum into primary spectra belonging to a first group and a secondary spectrum belonging to a second group; determining film thicknesses of the substrate from the primary spectra; and determining a film thickness at a measurement point corresponding to the secondary spectrum using the primary spectra or the film thicknesses.

A polishing method capable of measuring a film thickness of a substrate, such as a semiconductor wafer, having various structural elements on its surface with high accuracy is disclosed. The polishing method includes: generating spectra of reflected lights from measurement points on a substrate; classifying the spectra based on a shape of each spectrum into primary spectra belonging to a first group and a secondary spectrum belonging to a second group; determining film thicknesses of the substrate from the primary spectra; and determining a film thickness at a measurement point corresponding to the secondary spectrum using the primary spectra or the film thicknesses.

A method of training a neural network includes obtaining two ground truth thickness profiles a test substrate, obtaining two thickness profiles for the test substrate as measured by an in-situ monitoring system while the test substrate is on polishing pads of different thicknesses, generating an estimated thickness profile for another thickness value that is between the two thickness values by interpolating between the two profiles, and training a neural network using the estimated thickness profile.

A method of training a neural network includes obtaining two ground truth thickness profiles a test substrate, obtaining two thickness profiles for the test substrate as measured by an in-situ monitoring system while the test substrate is on polishing pads of different thicknesses, generating an estimated thickness profile for another thickness value that is between the two thickness values by interpolating between the two profiles, and training a neural network using the estimated thickness profile.

A neural network is trained for use in a substrate thickness measurement system by obtaining ground truth thickness measurements of a top layer of a calibration substrate at a plurality of locations, each location at a defined position for a die being fabricated on the substrate. A plurality of color images of the calibration substrate are obtained, each color image corresponding to a region for a die being fabricated on the substrate. A neural network is trained to convert color images of die regions from an in-line substrate imager to thickness measurements for the top layer in the die region. The training is performed using training data that includes the plurality of color images and ground truth thickness measurements with each respective color image paired with a ground truth thickness measurement for the die region associated with the respective color image.

A neural network is trained for use in a substrate thickness measurement system by obtaining ground truth thickness measurements of a top layer of a calibration substrate at a plurality of locations, each location at a defined position for a die being fabricated on the substrate. A plurality of color images of the calibration substrate are obtained, each color image corresponding to a region for a die being fabricated on the substrate. A neural network is trained to convert color images of die regions from an in-line substrate imager to thickness measurements for the top layer in the die region. The training is performed using training data that includes the plurality of color images and ground truth thickness measurements with each respective color image paired with a ground truth thickness measurement for the die region associated with the respective color image.