Method and apparatus for deciding cause of abnormality in plasma processing apparatus

1. A method of deciding a cause of abnormality in a plasma processing apparatus which performs a plasma processing to an object to be processed in a processing chamber, comprising:

an analysis data acquiring step of acquiring analysis data constituted by a plurality of parameters on the basis of detection values obtained in processes for the object from a detector arranged in the plasma processing apparatus;

an abnormality deciding step of analyzing the acquired analysis data to decide whether the analysis data is abnormal or not;

a degree-of-influence calculating step of calculating degrees of influence on the abnormality of each parameter of the analysis data decided as abnormal data; and

a cause-of-abnormality deciding step of deciding whether the analysis data is abnormal or not after removing the influences on the abnormality from the parameter sequentially in the descending order of the degrees of influence of the parameters, and then deciding the parameters from which the influences on the abnormality are removed up to now as parameters which cause the abnormality when it is decided that the analysis data is normal.

2. The method of deciding a cause of abnormality in a plasma processing apparatus according to claim 1 , wherein

the abnormality decision step has the step of performing a principal component analysis to the acquired analysis data to calculate a residual score, and the step of deciding the analysis data having the residual score which is exceeds a predetermined value as abnormal data,

the degree-of-influence calculating step calculates contributions to the residual score as degrees of influence on abnormality with respect to each parameter of the analysis data decided as abnormal data, and

the cause-of-abnormality deciding step sets contributions of the parameters at 0 or a value close to 0 in a descending order of contribution of the parameters to sequentially calculate residual scores, and, when the residual scores are not more than the predetermined value, decides the parameters having the contributions which are set at 0 or a value close to 0 until now as parameters which cause abnormality.

3. The method of deciding a cause of abnormality in a plasma processing apparatus according to claim 2 , wherein, in the cause-of-abnormality deciding step, in order to set the contribution of a specific parameter at 0 or a value close to 0, a correlation between the specific parameter and other parameters is calculated by a multivariate analysis, a prediction value of the specific parameter is calculated on the basis of the correlation, and the value of the specific parameter is replaced with the prediction value.

4. The method of deciding a cause of abnormality in a plasma processing apparatus according to claim 3 , wherein, the multivariate analysis is a partial least squares method.

5. An apparatus for deciding a cause of abnormality in a plasma processing apparatus which performs a plasma processing to an object to be processed in a processing chamber, comprising:

analysis data acquiring means for acquiring analysis data constituted by a plurality of parameters on the basis of detection values obtained in processes for the object from a detector arranged in the plasma processing apparatus;

abnormality deciding means for analyzing the acquired analysis data to decide whether the analysis data is abnormal or not;

degree-of-influence calculating means for calculating degrees of influence on the abnormality of each parameter of the analysis data decided as abnormal data; and

cause-of-abnormality deciding means for deciding whether the analysis data is abnormal or not after removing the influences on the abnormality from the parameter sequentially in the descending order of the degrees of influence of the parameters, and then deciding the parameters from which the influences on the abnormality are removed up to now as parameters which cause the abnormality when it is decided that the analysis data is normal.

6. The apparatus of deciding a cause of abnormality in a plasma processing apparatus according to claim 5 , wherein

the abnormality decision means performs a principal component analysis to the acquired analysis data to calculate a residual score and decides the analysis data having the residual score which is exceeds a predetermined value as abnormal data,

the degree-of-influence calculating means calculates contributions to the residual score as degrees of influence on abnormality with respect to each parameter of the analysis data decided as abnormal data, and

the cause-of-abnormality deciding means sets contributions of the parameters at 0 or a value close to 0 in a descending order of contribution of the parameters to sequentially calculate residual scores, and, when the residual scores are not more than the predetermined value, decides the parameters having the contributions which are set at 0 or a value close to 0 until now as parameters which cause abnormality.

7. The apparatus of deciding a cause of abnormality in a plasma processing apparatus according to claim 6 , wherein, in the cause-of-abnormality deciding means, in order to set the contribution of a specific parameter at 0 or a value close to 0, a correlation between the specific parameter and other parameters is calculated by a multivariate analysis, a prediction value of the specific parameter is calculated on the basis of the correlation, and the value of the specific parameter is replaced with the prediction value.

8. The apparatus of deciding a cause of abnormality in a plasma processing apparatus according to claim 7 , wherein, the multivariate analysis is a partial least squares method.