Technique for training neural network for use in in-situ monitoring during polishing and polishing system

Patent metadata

Field	Value
Publication number	US-12136574-B2
Application number	US-202318471893-A
Country	US
Kind code	B2
Filing date	Sep 21, 2023
Priority date	May 14, 2020
Publication date	Nov 5, 2024
Grant date	Nov 5, 2024

Abstract

A method of polishing a substrate includes polishing a conductive layer on the substrate at a polishing station, monitoring the layer with an in-situ eddy current monitoring system to generate a plurality of measured signals values for a plurality of different locations on the layer, generating thickness measurements the locations, and detecting a polishing endpoint or modifying a polishing parameter based on the thickness measurements. The conductive layer is formed of a first material having a first conductivity. Generating includes calculating initial thickness values based on the plurality of measured signals values and processing the initial thickness values through a neural network that was trained using training data acquired by measuring calibration substrates having a conductive layer formed of a second material having a second conductivity that is lower than the first conductivity to generated adjusted thickness values.

First claim

What is claimed is:

1. A computer program product, tangibly embodied in a non-transitory computer readable medium, comprising instructions to cause one or more computers to:
receive from a monitoring system a first set of sensor measurements generated by scanning a sensor of an in-situ monitoring system across a calibration substrate having a conductive layer formed of a first material having a first conductivity;
receive ground truth measurements of a thickness of the conductive layer of the calibrations substrate to provide a first thickness profile;
scale the first thickness profile based the first conductivity and a target second conductivity that is greater than the first conductivity to provide a modified second thickness profile equivalent to a thickness profile that would be generated if the conductive layer were formed of a second material of the second conductivity;
train a neutral network to convert sensor measurements from the in-situ monitoring system to thickness measurements for a layer formed of the second material to generate a trained neural network, the training performed using training data including the modified second thickness profile and calibration thickness values based on the first set of sensor measurements from the conductive layer formed of the first material;
during polishing of a conductive layer formed of a first material having a first conductivity on a substrate at a polishing station, receive from an in-situ eddy current monitoring system a plurality of measured signals values for a plurality of different locations on the layer;
generate thickness measurements for the locations, the instructions to generate thickness measurements including instructions to calculate initial thickness values based on the plurality of measured signals values and process the initial thickness values through the neural network; and
at least one of detect a polishing endpoint or modify a polishing parameter based on the thickness measurements.

2. The computer program product of claim 1 , wherein the instructions to train include instructions to train the neural network using training data acquired by measuring calibration substrates having a conductive layer formed of the first material.

3. The computer program product of claim 1 , wherein the instructions to calculate initial thickness values include instructions to convert the measured signal values to initial thickness values using a correlation curve.

4. The computer program product of claim 3 , comprising instructions to input the initial thickness values into input nodes of the neural network and at least some of the thickness measurements from output nodes of the neural network.

5. The computer program product of claim 4 , comprising instructions to, for each location of a second plurality of different locations, receive a signal value from the sensor of the in-situ eddy current monitoring system and convert the signal value to an initial thickness value using the correlation curve, and wherein the initial thickness value provides one of the thickness measurements without correcting the initial thickness value using the neural network.

6. The computer program product of claim 1 , wherein the instructions to scale the first thickness profile comprise instructions to divide ground truth measurements by a ratio of the second conductivity to the first conductivity.

7. A method of polishing a substrate, comprising:
scanning a sensor of an in-situ monitoring system across a calibration substrate having a conductive layer formed of a first material having a first conductivity, the monitoring system generating a first set of sensor measurements;
obtaining ground truth measurements of a thickness of the conductive layer of the calibrations substrate to provide a first thickness profile;
scaling the first thickness profile based the first conductivity and a target second conductivity that is greater than the first conductivity to provide a modified second thickness profile equivalent to a thickness profile that would be generated if the conductive layer were formed of a second material of the second conductivity,
training a neutral network to convert sensor measurements from the in-situ monitoring system to thickness measurements for a layer formed of the second material to generate a trained neural network, the training performed using training data including the modified second thickness profile and calibration thickness values based on the first set of sensor measurements from the conductive layer formed of the first material;
during polishing of a conductive layer on a substrate at a polishing station, the conductive layer formed of a first material having a first conductivity;
monitoring the layer during polishing at the polishing station with an in-situ eddy current monitoring system to generate a plurality of measured signals values for a plurality of different locations on the layer;
generating thickness measurements for the locations, the generating including calculating initial thickness values based on the plurality of measured signals values and processing the initial thickness values through the trained neural network; and
at least one of detecting a polishing endpoint or modifying a polishing parameter based on thickness measurements.

8. The method of claim 7 , comprising training the neural network using training data acquired by measuring calibration substrates having a conductive layer formed of the first material.

9. The method of claim 7 , wherein the first material and the second material are selected from the group including copper, aluminum, cobalt, tungsten, titanium and titanium nitride.

10. The method of claim 9 , wherein the first material is copper and the second material is tungsten or titanium nitride.

11. The method of claim 7 , wherein obtaining ground truth measurements comprises measuring the calibration substrate with a four-point probe.

12. A polishing system comprising:
a platen to support a polishing pad;
a carrier head to hold a substrate in contact with the polishing pad;
an in-situ eddy current monitoring system; and
a controller configured to
receive from the monitoring system a first set of sensor measurements generated by scanning a sensor of an in-situ monitoring system across a calibration substrate having a conductive layer formed of a first material having a first conductivity,
receive ground truth measurements of a thickness of the conductive layer of the calibrations substrate to provide a first thickness profile,
scale the first thickness profile based the first conductivity and a target second conductivity that is greater than the first conductivity to provide a modified second thickness profile equivalent to a thickness profile that would be generated if the conductive layer were formed of a second material of the second conductivity,
train a neutral network to convert sensor measurements from the in-situ monitoring system to thickness measurements for a layer formed of the second material to generate a trained neural network, the training performed using training data including the modified training profile and calibration thickness values based on the first set of sensor measurements from the conductive layer formed of the first material,
receive from the in-situ eddy current monitoring system a plurality of measured signals values for a plurality of different locations on the layer;
generate thickness measurements for the locations by calculating initial thickness values based on the plurality of measured signals values and processing the initial thickness values through the trained neural network, and
at least one of detect a polishing endpoint

Assignees

• Applied Materials Inc

Inventors

• Xu Kun
• Shrestha Kiran Lall
• Bennett Doyle E
• Gage David Maxwell
• Cherian Benjamin
• Qian Jun
• Lee Harry Q

Classifications

• H10P74/238

  comprising acting in response to an ongoing measurement without interruption of processing, e.g. endpoint detection or in-situ thickness measurement · CPC title

• H10P74/203 Primary

  Structural properties, e.g. testing or measuring thicknesses, line widths, warpage, bond strengths or physical defects · CPC title

• H10P52/403

  of conductive or resistive materials · CPC title

• H10P52/00

  Grinding, lapping or polishing of wafers, substrates or parts of devices · CPC title

• H10P74/207 Primary

  Electrical properties, e.g. testing or measuring of resistance, deep levels or capacitance-voltage characteristics · CPC title