A semiconductor chip manufacturer used to rely on manual labor to do sample inspection on film thickness, in which anomaly analysis was difficulty due to different production batches. Large amounts of data on production equipment were however not used. Due to a lack of effective analysing methods on film thickness and defects in CMP process, there was a high defect rate.
With its integrated solutions, GETECH has helped the enterprise to realize quality optimization of CMP&TF processes. All the parameters that relate with equipment and processes are imported to the system, and with automated feature extraction algorithms such as dimensional reduction and expansion, the data are selected and used to create factor collections that can be modified and trained by users. CMP film thickness anomaly and defect cause analysis models are created. Different kinds of complex data that affect film thickness are applied in the modeling that based on complex algorithms, thus predicting film thickness precisely. The anomalies were intercepted promptly, and their properties were “fully inspected in real-time” which has reduced the sampling inspection frequency by 60%. Our solution has reduced the defect rate of CMP related products by 70% with average annual economic benefits of over 10 million yuan.