During the recent years, we have seen lot of recalls from various big automotive industries. This really hits the customer satisfaction and product value in the market, which will impact the revenue of the company, and will also loose the edge over the competitors in the market. One of the foremost things for the semiconductor industry, which supplies chips that are placed in brakes, or in other electronic controls in the Automotive, should be maintained at very high quality. So, the industry spends a lot of their revenue in maintaining the quality of the product and trying to identify the issues related to the manufacturing of the product. If the industry is able to detect the quality issues during the Fab process, it will save them a lot of time, as well as the cost in manufacturing the finished product, and then scrapping it due to not meeting the quality standards.
Due to increase in the quality standards of the product today, the industry is spending huge amount in capturing the large volumes of data during the Procurement Process, Order Fulfilment Process, Manufacturing Process and Final Distribution Process to the end customer. This data is lying in Silos in the system. The systems are not integrated from end-to-end. Having the integrated system from Procurement of raw wafers, to Distribution process of the finished product, to the end customer, will help to see the full picture of the movement of order from start to end. Every manufacturing organization has the process of capturing the Manufacturing data (Fab to Final Test), for example, at Part Level and Lot Level, where the data is stored in the local system.
How this data is used currently:
The industry uses this stored data when a particular quality issue is reported by the customer (which is linked from a raw wafer to a finished product, which is shipped to the end customer). Using this link, the Quality Engineer identifies all the Lots associated with the raw wafer, determines the customers to whom they are shipped to (if shipped), determines where they are currently in the manufacturing process and distribution process, and then quarantine the lots, so that they will not be used for further manufacturing process and then they will scrap them. The disadvantage of this process is to take action in a reactive mode, after receiving the customer quality incident; this will hit the customer trust on the industry. The next thing, the industry needs to adapt to, is to use the advanced analytics, which can help them to contain the defect lots before they ship to the end customer, or move to the next stage of completion (like move from FAB -> ASY -> TEST -> DISTRIBUTION SYSTEM).
How Advanced Analytics can help to improve the Quality Defect Analytics:
In order to apply analytics to extract the outcome of the data, we need to build a system, which can link the data from Raw Wafer -> Finished Product -> Package, which is shipped to the end customer. This process of linking the data from the beginning of the process to the end of process is called Traceability. This can be done in different ways based on the business processes of the organization. Some organizations uses the Lot to do the linking from start to end, which they call as Lot Traceability. Some organizations trace their complete process using the Production order, which is placed by the customer and then tied to the Process order created in the manufacturing process. At each step, Quality data (eg: Measurement Data and Equipment Data) is captured in the system, which will be used to analyze to come up with the Standard Quality Parametric Data. Now, this data becomes the data set for applying the predictive analytics with the help of the analytics, models will be built to measure the quality of the product which will be integrated with the Machine learning system. This will analyse the data continuously and will be able to provide alerts to the Shop Floor engineer at each step of the manufacturing, and will help them to contain the bad lots moving further. This will improve the Quality of the product delivered to the end customer.