Semiconductor crystal manufacturing equipment management method, semiconductor crystal manufacturing method, and semiconductor crystal manufacturing management system

摘要

PROBLEM TO BE SOLVED: To reduce quality variation of a semiconductor crystal between devices and batches by managing each physical property value of a constituent member of a semiconductor crystal manufacturing device. SOLUTION: In this management method, a heat transfer simulation is performed using a plurality of calculation conditions created by changing each physical property value of each constituent member within a range in which the temperature distribution in the device can be calculated, and the temperature distribution in the device is calculated. The calculation result of the temperature distribution in the target area for suppressing the variation of the temperature distribution and the multiple data sets consisting of multiple calculation conditions are used as training data, the input is each physical property value of each component, and the output is the temperature of the target area. The first estimation model to be the distribution is created by the machine learning method, each physical property value of each component is changed within the range that can vary, and it is input to the first estimation model, the temperature distribution of the target area is estimated, and the estimated target. The permissible range of each physical property value of each component is calculated based on the temperature distribution of the region and the temperature distribution of the target target area, and each configuration is compared with each measured value and the permissible range of each physical property value of each component. Decide whether or not to replace the parts. [Selection diagram] FIG. 5