Predicting equipment fail mode from process trace

1. A method, comprising:

receiving into a computer-based machine learning model equipment trace data from a plurality of semiconductor equipment sensors during a plurality of steps in a semiconductor process;

detecting by the machine learning model a first anomaly in the equipment trace data, the first anomaly having an associated location within the equipment trace data;

defining by the machine learning model a window containing a period of time of the equipment trace data including the first anomaly;

calculating by the machine learning model statistics on the period of the equipment trace data in the window;

storing in memory the calculated statistics and the associated location of the first anomaly as a plurality of key features associated with the first anomaly;

searching through a database of past trace data by providing the plurality of key features of the first anomaly as inputs to the machine learning model configured to find past trace data having the plurality of key features;

determining by the machine learning model that an instance of the past trace data has the plurality of key features of the first anomaly;

identifying by the machine learning model a root cause for the instance of the past trace data having the key features of the first anomaly; and

taking an action to correct the root cause in the semiconductor process.

2. The method of claim 1 , further comprising:

retrieving the root cause and a corrective action for the root cause from the database.

3. The method of claim 1 , wherein the determining step further comprises:

determining by the machine learning model a likelihood that the instance of the past trace data in the database has the key features of the first anomaly; and

retrieving the root cause if the likelihood exceeds a threshold.

4. The method of claim 3 , wherein the step of retrieving a root cause further comprises:

retrieving a corrective action for the root cause from the database.

5. A method for predicting semiconductor processing equipment failure, comprising:

detecting, by a processor including a machine-learning model trained to detect anomalies in sets of trace data using multivariate analysis, a first anomalous pattern in a first set of traces obtained from a plurality of semiconductor equipment sensors during a plurality of steps in a semiconductor process;

identifying, by the processor, a time-period window that contains the first anomalous pattern in the first set of traces;

calculating, by the processor using multivariate analysis, a plurality of features from the first set of traces located within the window;

searching, by the processor, a database of past trace data;

identifying, by the processor, at least one set of past trace data in the database having the plurality of features in an associated anomalous pattern;

determining, by the processor using multivariate analysis of the plurality of features in the associated anomalous pattern, a likelihood that the associated anomalous pattern of the at least one set of past trace data is the same as the first anomalous pattern;

retrieving, by the processor, a root cause for the at least one prior anomalous pattern from the database if the likelihood exceeds a threshold; and

taking an action to correct the root cause in the semiconductor process.

6. The method of claim 5 , the step of identifying a time-period window further comprising:

defining the time-period window as a region where the values in the first set of traces are changing rapidly.

7. The method of claim 5 , the step of identifying a time-period window further comprising:

defining the time-period window as a region where the rate of change of values in the first set of traces is changing rapidly.