OPTIMIZING SEMICONDUCTOR MANUFACTURING PROCESSES USING MACHINE LEARNING
In some embodiments, a computer-implemented method of controlling a semiconductor manufacturing process is provided. A computing system generates predicted metrology values for a current run and a next run by providing metrology forecast inputs to a metrology forecast model. The computing system generates an updated recipe for executing at least one semiconductor manufacturing process step using the predicted metrology values for the current run and the next run.
1 . (canceled)
2 . A method of controlling a semiconductor manufacturing process, the method comprising:
determining inputs, comprising one or both of sensor data and a process recipe from a manufacturing equipment, for a metrology forecast model;
generating, by inputting the inputs into the metrology forecast model, predicted metrology values for a current manufacturing run and predicted metrology values for a next manufacturing run subsequent to the current manufacturing run; and
generating process inputs based on the process recipe for the manufacturing equipment based on the predicted metrology values for the current manufacturing run and the predicted metrology values for the next manufacturing run.
3 . The method of claim 2 , wherein the manufacturing equipment is configured to perform at least one of thin film deposition, photolithography, etching, overlay correction, or chemical mechanical planarization.
4 . The method of claim 2 , wherein the inputs include one or more of process input values, trace statistic values, exogenous values, apriori values, or measured metrology values.
5 . The method of claim 2 ,
wherein generating the process inputs comprises providing the predicted metrology values for the current manufacturing run and the predicted metrology values for the next manufacturing run to an actor model; and
wherein the method further comprises retraining alpha parameters of the actor model after a predetermined period of time or a predetermined number of runs.
6 . The method of claim 2 , wherein generating the process inputs comprises:
providing process model inputs to a process model to determine predicted process outputs; and
evaluating the predicted process outputs and the predicted metrology values for the current manufacturing run and the predicted metrology values for the next manufacturing run using a cost function to determine the process inputs.
7 . The method of claim 6 , wherein the process model inputs include one or more of a deposition time value, a high frequency (HF) power value, an argon flow value, a pedestal gap value, a dosing value, an etch time value, or an etch gas flow value.
8 . The method of claim 6 , wherein the process model is linearized about an operating point in a space of the process model inputs.
9 . The method of claim 6 , wherein output of the process model includes a prediction for each output dimension within a space of the process model inputs.
10 . The method of claim 6 , further comprising retraining the process model in response to determining that a variance in an independent input space exceeds a threshold variance proportional to known model parameter uncertainty.
11 . The method of claim 2 , further comprising retraining the metrology forecast model in response to obtaining measured metrology values.
12 . Non-transitory computer readable storage media storing instructions that when executed by a system of one or more processors, cause the one or more processors to:
determine inputs, comprising one or both of sensor data and a process recipe from a manufacturing equipment, for a metrology forecast model;
generate, by inputting the inputs into the metrology forecast model, predicted metrology values for a current manufacturing run and predicted metrology values for a next manufacturing run subsequent to the current manufacturing run; and
generate process inputs based on the process recipe for the manufacturing equipment based on the predicted metrology values for the current manufacturing run and the predicted metrology values for the next manufacturing run.
13 . The non-transitory computer readable storage media of claim 12 , wherein the manufacturing equipment is configured to perform at least one of thin film deposition, photolithography, etching, overlay correction, or chemical mechanical planarization.
14 . The non-transitory computer readable storage media of claim 12 , wherein the inputs include one or more of process input values, trace statistic values, exogenous values, apriori values, or measured metrology values.
15 . The non-transitory computer readable storage media of claim 12 ,
wherein to generate the process inputs the instructions cause the one or more processors to provide the predicted metrology values for the current manufacturing run and the predicted metrology values for the next manufacturing run to an actor model; and
wherein the instructions further cause the one or more processors to retrain alpha parameters of the actor model after a predetermined period of time or a predetermined number of runs.
16 . The non-transitory computer readable storage media of claim 12 , wherein to generate the process inputs the instructions cause the one or more processors to
provide process model inputs to a process model to determine predicted process outputs; and
evaluate the predicted process outputs and the predicted metrology values for the current manufacturing run and the predicted metrology values for the next manufacturing run using a cost function to determine the process inputs.
17 . The non-transitory computer readable storage media of claim 16 , wherein the process model inputs include one or more of a deposition time value, a high frequency (HF) power value, an argon flow value, a pedestal gap value, a dosing value, an etch time value, or an etch gas flow value.
18 . The non-transitory computer readable storage media of claim 16 , wherein the process model is linearized about an operating point in a space of the process model inputs.
19 . The non-transitory computer readable storage media of claim 16 , wherein output of the process model includes a prediction for each output dimension within a space of the process model inputs.
20 . The non-transitory computer readable storage media of claim 16 , wherein the instructions further cause the one or more processors to retrain the process model in response to determining that a variance in an independent input space exceeds a threshold variance proportional to known model parameter uncertainty.
21 . The non-transitory computer readable storage media of claim 12 , wherein the instructions further cause the one or more processors to retrain the metrology forecast model in response to obtaining measured metrology values.