State prediction apparatus and semiconductor manufacturing apparatus

1. A state prediction apparatus that predicts a state of a plasma processing apparatus, comprising:

a computer configured to

acquire training data comprising a data group for a plurality of samples, the data group including extracted feature data obtained in a normal etching apparatus state;

calculate a first set of features that indicates a state of the plasma processing apparatus based on the acquired training data of the plasma processing apparatus when the state of the plasma processing apparatus is a normal state;

acquire a second set of features comprising testing data for detecting an abnormality;

standardize the features in the second set using the features in the first set, wherein the standardized features of the second set comprise a value indicating a degree of deviation of the plasma processing apparatus from the normal state in the second set of features;

create a ranking of the features of the standardized features of the second set;

select a number in descending order of the ranked features;

create a model that predicts the state of the plasma processing apparatus using a subset of the first set of features, which is composed of a same kind of features selected in descending order of the standardized features in the second set; and

predict the state of the plasma processing apparatus by using the generated model.

2. The state prediction apparatus according to claim 1 , wherein

the generated model is regenerated each time the state of the plasma processing apparatus is predicted.

3. The state prediction apparatus according to claim 1 , wherein

the standardized features in the second set are each a value obtained by dividing a value, which is obtained by subtracting an average value of the features in the first set in a plurality of plasma processings from the features in the second set, by a standard deviation of the features in the first set in the plurality of plasma processings.

4. A semiconductor manufacturing apparatus, comprising:

a processing chamber in which a sample is processed;

a transfer chamber that transfers the sample to the processing chamber and is connected to the processing chamber; and

a control device comprising a computer configured to

acquire training data comprising a data group for a plurality of samples, the data group including feature data extracted in a normal etching apparatus state;

calculate a first set of features that indicates a state of the plasma processing apparatus based on the acquired training data of the plasma processing apparatus when the state of the plasma processing apparatus is a normal state;

acquire a second set of features comprising testing data for detecting an abnormality;

standardize the features in the second set using the features in the first set, wherein the standardized features of the second set comprise a value indicating a degree of deviation of the plasma processing apparatus from the normal state in the second set of features;

create a ranking of the features of the standardized features of the second set;

select a number in descending order of the ranked features;

create a model that predicts the state of the plasma processing apparatus using a subset of the first set of features, which is composed of a same kind of features selected in descending order of the standardized features in the second set; and

predict the state of the plasma processing apparatus by using the generated model.

5. The semiconductor manufacturing apparatus according to claim 4 , wherein

any position in the semiconductor manufacturing apparatus is the processing chamber or the transfer chamber.

6. The semiconductor manufacturing apparatus according to claim 5 , wherein

the standardized features in the second set are each a value obtained by dividing a value, which is obtained by subtracting an average value of the features in the first set in a plurality of processings from the features in the second set, by a standard deviation of the features in the first set in the plurality of processings.

7. A semiconductor manufacturing apparatus comprising:

a processing chamber in which a sample is processed and a transfer chamber which transfers the sample to the processing chamber and is connected to the processing chamber; and

a state prediction apparatus comprising a computer configured to

acquire training data comprising a data group for a plurality of samples, the data group including feature data extracted in a normal etching apparatus state;

calculate a first set of features that indicates a state of the plasma processing apparatus based on the acquired training data of the plasma processing apparatus when the state of the plasma processing apparatus is a normal state;

acquire a second set of features comprising testing data for detecting an abnormality;

standardize the features in the second set using the features in the first set, wherein the standardized features of the second set comprise a value indicating a degree of deviation of the plasma processing apparatus from the normal state in the second set of features;

create a ranking of the features of the standardized features of the second set;

select a number in descending order of the ranked features;

create a model that predicts the state of the plasma processing apparatus using a subset of the first set of features, which is composed of a same kind of features selected in descending order of the standardized features in the second set; and

predict the state of the plasma processing apparatus by using the generated model.

8. The semiconductor manufacturing apparatus according to claim 7 , wherein

any position in the semiconductor manufacturing apparatus is the processing chamber or the transfer chamber.

9. The semiconductor manufacturing apparatus according to claim 8 , wherein

the standardized features in the second set are each a value obtained by dividing a value, which is obtained by subtracting an average value of the features in the first set in a plurality of processings from the features in the second set, by a standard deviation of the features in the first set in the plurality of processings.