Semiconductor process prediction method and semiconductor process prediction apparatus considering overall features and local features

1 . A semiconductor process prediction method considering overall features and local features, comprising:

obtaining a plurality of equipment sensing curves;

filtering the equipment sensing curves to reduce a colinearity of the equipment sensing curves, wherein the step of filtering the equipment sensing curves comprises:

classifying the equipment sensing curves into a plurality of groups according to a correlation matrix; and

selecting one from the equipment sensing curves in each of the groups;

performing, by an aligning circuit, a dynamic time warping (DTW) to align the equipment sensing curves with a template curve, wherein the equipment sensing curves and the template curve are continuous curves;

inputting the equipment sensing curves which are aligned into a Convolutional Neural Network (CNN) model, to obtain a first prediction result considering the local features;

performing a statistical analysis procedure on the equipment sensing curves to obtain a plurality of statistical data;

inputting the statistical data into an Artificial Neural Network (ANN) model, to obtain a second prediction result considering the overall features; and

obtaining a total prediction result according to the first prediction result and the second prediction result;

wherein the first prediction result considering the local features and the second prediction result considering the overall features are obtained based on identical production lot;

wherein the equipment sensing curves are used to obtain a plurality of continuous curve features including bursts, drift and oscillations and the statistical data including mean, standard deviation and median, and the continuous curve features and the statistical data of the equipment sensing curves are inputted into the CNN model and the ANN model respectively;

wherein the CNN model and the ANN model are implemented by at least one circuit or chip to process the equipment sensing curves and the statistical data, respectively.

2 . The semiconductor process prediction method considering the overall features and the local features according to claim 1 , wherein each of the equipment sensing curves is composed of sensing data continuously captured over time.

3 . The semiconductor process prediction method considering the overall features and the local features according to claim 1 , wherein the CNN model is a single-channel model.

4 . The semiconductor process prediction method considering the overall features and the local features according to claim 1 , wherein the CNN model is a multi-channel model.

5 . The semiconductor process prediction method considering the overall features and the local features according to claim 1 , wherein in the step of obtaining the total prediction result according to the first prediction result and the second prediction result, the total prediction result is obtained through a voting procedure.

6 . The semiconductor process prediction method considering the overall features and the local features according to claim 1 , wherein the CNN model is a LeNet model, an AlexNet model, a VGG model, a GoogLeNet model or a ResNet model.

7 . The semiconductor process prediction method considering the overall features and the local features according to claim 1 , wherein each of the statistical data is mean, standard deviation or median.

8 . The semiconductor process prediction method considering the overall features and the local features according to claim 1 , wherein the ANN model is a Supervised Learning Network, an Unsupervised Learning Network, a Hybrid Learning Network, an Associate Learning Network or an Optimization Application Network.

9 . A semiconductor process prediction apparatus considering overall features and local features, comprising:

a database, configured to store a plurality of equipment sensing curves;

a filtering unit, configured to filter the equipment sensing curves to reduce a colinearity of the equipment sensing curves, wherein the filtering unit classifies the equipment sensing curves into a plurality of groups according to a correlation matrix, and selects one from the equipment sensing curves in each of the groups;

an aligning circuit, configured to perform a dynamic time warping (DTW) to align the equipment sensing curves with a template curve, wherein the equipment sensing curves and the template curve are continuous curves;

a Convolutional Neural Network (CNN) model, configured to receive the equipment sensing curves which are aligned to obtain a first prediction result considering the local features;

a statistical unit, configured to perform a statistical analysis procedure on the equipment sensing curves to obtain a plurality of statistical data;

an Artificial Neural Network (ANN) model, configured to receive the statistical data to obtain a second prediction result considering the overall features; and

a total prediction unit, configured to obtain a total prediction result according to the first prediction result and the second prediction result;

wherein the first prediction result considering the local features and the second prediction result considering the overall features are obtained based on identical production lot;

wherein the equipment sensing curves are used to obtain a plurality of continuous curve features including bursts, drift and oscillations and the statistical data including mean, standard deviation and median, and the continuous curve features and the statistical data of the equipment sensing curves are inputted into the CNN model and the ANN model respectively;

wherein the CNN model and the ANN model are implemented by at least one circuit or chip to process the equipment sensing curves and the statistical data, respectively.

10 . The semiconductor process prediction apparatus considering the overall features and the local features according to claim 9 , wherein each of the equipment sensing curves is composed of sensing data continuously captured over time.

11 . The semiconductor process prediction apparatus considering the overall features and the local features according to claim 9 , wherein the CNN model is a single-channel model.

12 . The semiconductor process prediction apparatus considering the overall features and the local features according to claim 9 , wherein the CNN model is a multi-channel model.

13 . The semiconductor process prediction apparatus considering the overall features and the local features according to claim 9 , wherein the total prediction unit obtains the total prediction result through a voting procedure.

14 . The semiconductor process prediction apparatus considering the overall features and the local features according to claim 9 , wherein the CNN model is a LeNet model, an AlexNet model, a VGG model, a GoogleNet model or a ResNet model.

15 . The semiconductor process prediction apparatus considering the overall features and the local features according to claim 9 , wherein each of the statistical data is mean, standard deviation or median.

16 . The semiconductor process prediction apparatus considering the overall features and the local features according to claim 9 , wherein the ANN model is a Supervised Learning Network, an Unsupervised Learning Network, a Hybrid Learning Network, an Associate Learning Network or an Optimization Application Network.