Polishing apparatus using machine learning and compensation for pad thickness
Data received from an in-situ monitoring system includes, for each scan of a sensor, a plurality of measured signal values for a plurality of different locations on a layer. A thickness of a polishing pad is determined based on the data from the in-situ monitoring system. For each scan, a portion of the measured signal values are adjusted based on the thickness of the polishing pad. For each scan of the plurality of scans and each location of the plurality of different locations, a value is generated representing a thickness of the layer at the location. This includes processing the adjusted signal values using one or more processors configured by machine learning. A polishing endpoint is detected or a polishing parameter is modified based on the values representing the thicknesses at the plurality of different locations.
1. A computer storage medium encoded with instructions that, when executed by one or more computers, cause the one or more computers to perform operations comprising:
receiving data from an in-situ eddy current monitoring system, the data including, for each scan of a plurality of scans of a sensor of the in-situ eddy current monitoring system of a conductive layer of a substrate being polished by a polishing pad, a plurality of measured signal values for a plurality of different locations on the conductive layer;
determining a measured thickness of the polishing pad based on the data from the in-situ eddy current monitoring system,
for each scan of the plurality of scans, generating, for each location of the plurality of different locations, a value representing a thickness of the conductive layer at the location, thereby providing a plurality of values representing thicknesses at the plurality of different locations, wherein the generating includes processing at least the plurality of measured signal values and the measured thickness of the polishing pad using one or more processors configured by machine learning, and
at least one of detecting a polishing endpoint or modifying a polishing parameter based on the plurality of values representing the thicknesses at the plurality of different locations.
2. The computer storage medium of claim 1 , wherein processing at least the plurality of measured signal values and the measured thickness of the polishing pad includes applying at least some of the plurality of measured signal values to a plurality of first input nodes of a trained neural network and applying the measured thickness of the polishing pad to a second input node of the trained neural network.
3. The computer storage medium of claim 2 , wherein the operations comprise applying a group of measured signal values which is less than all of the plurality of measured signal values to the plurality of first input nodes.
4. The computer storage medium of claim 3 , wherein the plurality of measured signal values includes a first group of values from locations in an edge region of the substrate and a second group of values from locations in a center region of the substrate.
5. The computer storage medium of claim 4 , wherein the operations comprise applying the first group of values to the plurality of first input nodes.
6. The computer storage medium of claim 5 , wherein the second group of values are not applied to the neural network.
7. The computer storage medium of claim 1 , wherein determining the thickness of the polishing pad includes determining a gain for the substrate and determining the thickness of the polishing pad from a correlation function that provides pad thickness values as a function of gain.
8. The computer storage medium of claim 7 , wherein determining the gain for the substrate includes determining an estimated starting thickness value of the layer based on the plurality of values representing thicknesses for a plurality of scans occurring in an initial portion of a polishing operation of the layer, receiving an initial thickness value of the layer prior to polishing, and comparing the estimated starting thickness value to the received initial thickness value.
9. The computer storage medium of claim 8 , wherein the operations comprise receiving a plurality of raw thickness values from the sensor, and wherein the operations include, after determining the gain, multiplying the plurality of raw thickness values by the gain to generate the plurality of measured signal values.
10. The computer storage medium claim 1 , comprising adjusting at least a portion of the plurality of measured signal values by performing a convolution on the at least the portion of the plurality of measured signal values.
11. The computer storage medium of claim 1 , wherein the plurality of different locations on the layer include at least one edge location in an edge region of the substrate, at least one central location in a central region of the substrate, and at least one anchor location in an anchor region between the central region and the edge region.
12. The computer storage medium of claim 11 , wherein the at least a portion of the plurality of measured signal values includes measured signal values corresponding to the edge region, includes measured signal values corresponding to the anchor region, and does not includes measured signal values corresponding to the central region.
13. The computer storage medium of claim 1 , wherein generating the plurality of values representing thicknesses uses a correlation curve that provides layer thickness value as a function of signal value.
14. A polishing system, comprising:
a support for a polishing pad;
a carrier to hold a substrate in contact with the polishing pad;
an in-situ eddy current monitoring system having a sensor;
a motor to generate relative motion between the sensor and the substrate such that the sensor makes a plurality of scans across the substrate, the in-situ eddy current monitoring system configured to generate data including, for each scan of the plurality of scans, a plurality of measured signal values for a plurality of different locations on a conductive layer of the substrate; and
a controller configured to
receive data from the in-situ eddy current monitoring system, the data including, for each scan of a plurality of scans of the sensor, a plurality of measured signal values for a plurality of different locations on the conductive layer;
determine a measured thickness of the polishing pad based on the data from the in-situ eddy current monitoring system,
for each scan of the plurality of scans, generate, for each location of the plurality of different locations, a value representing a thickness of the conductive layer at the location, thereby providing a plurality of values representing thicknesses at the plurality of different locations, wherein the controller is configured to process at least the plurality of measured signal values and the measured thickness of the polishing pad using one or more processors configured by machine learning to generate the value representing the thickness of the conductive layer, and
at least one of detect a polishing endpoint or modify a polishing parameter based on the plurality of values representing the thicknesses at the plurality of different locations.
15. The system of claim 14 , wherein the controller is configured to process at least the plurality of measured signal values and the measured thickness of the polishing pad by applying at least some of the plurality of measured signal values to a plurality of first input nodes of a trained neural network and applying the measured thickness of the polishing pad to a second input node of the trained neural network.
16. The system of claim 15 , wherein the controller is configure to apply a group of measured signal values which is less than all of the plurality of measured signal values to the plurality of first input nodes.
17. A method of polishing a substrate, comprising:
bringing the substrate into contact with a polishing pad at a polishing station;
moving the substrate relative to the polishing pad to polish a layer on the substrate;
monitoring the layer during polishing at the polishing station with a sensor of an in-situ monitoring system, the sensor moving relative to the substrate to provide a plurality of scans of the sensor across the substrate, the in-situ monitoring system generating data including, for each scan of the plurality of scans, a plurality of measured signal values for a plurality of different locations on the layer;
determining a measured thickness of the polishing pad based on the data from the in-situ monitoring system,
for each scan of the plurality of scans, generating, for each location of the plurality of different locations, a value representing a thickness of the layer at the location, thereby providing a plurality of values representing thicknesses at the plurality of different locations, wherein the generating includes processing at least the plurality of measured signal values and the measured thickness of the polishing pad using one or more processors configured by machine learning, and
at least one of detecting a polishing endpoint or modifying a polishing parameter based on the plurality of values representing the thicknesses at the plurality of different locations.
18. The method of claim 17 , comprising processing at least the plurality of measured signal values and the measured thickness of the polishing pad by applying at least some of the plurality of measured signal values to a plurality of first input nodes of a trained neural network and applying the measured thickness of the polishing pad to a second input node of the trained neural network.
19. The method of claim 18 , comprising applying a group of measured signal values which is less than all of the plurality of measured signal values to the plurality of first input nodes.