| CPC B24B 37/013 (2013.01) [G05B 19/4188 (2013.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); H01L 23/00 (2013.01); G05B 2219/32335 (2013.01)] | 13 Claims |
|
1. A method of training a neural network, comprising:
for each test substrate of a plurality of test substrates having a different thickness profile, obtaining a ground truth thickness profile for the test substrate;
obtaining a first thickness value;
for each test substrate of the plurality of test substrates, obtaining a first measured thickness profile corresponding to the test substrate being measured by an in-situ monitoring system while on a polishing pad of a first thickness corresponding to the first thickness value;
obtaining a second thickness value;
for each test substrate of the plurality of test substrates, obtaining a second measured thickness profile corresponding to the test substrate being measured by the in-situ monitoring system while on a polishing pad of a second thickness corresponding to the second thickness value;
for each test substrate of the plurality of test substrates, generating an estimated third thickness profile for a third thickness value that is between the first thickness value and the second thickness value by interpolating between the first measured thickness profile for the test substrate and the second measured thickness profile; and
training a neural network that has a plurality of input nodes and a plurality of output nodes by, for each test substrate, applying the estimated third thickness profile to a multiplicity of input nodes from the plurality of input nodes, applying the third thickness value to an input node from the plurality of input nodes or to an intermediate node in the neural network, and applying the ground truth thickness profile to a plurality of output nodes while the neural network is in a training mode.
|