A standardized data warehouse helps identify yield issues faster.
In semiconductor manufacturing, huge amounts of data are generated and collected at every step in multiple production areas, with data coming from wafer fab, probe/testing, assembly, and final test. That data is usually stored separately in its respective manufacturing department, isolated from other departments.
In order to analyze the production data and make better decisions, yield and test engineers need the data to be integrated and merged into one database.
Siloed data hinders business operations and the data analysis initiatives intended to support them. Apart from being completely isolated from other departments, siloed data is often incompatible with other data sets or formats. A well-planned data management strategy is key in achieving data that is cleaned, parsed and integrated.
Fixing siloed data in semiconductor manufacturing results in:
High-volume data coming from many different sources in manufacturing and test (such as inline data, parameter test, wafer sort and final test) arrives in many different data formats, all of which must be combined and consolidated.
By using patented algorithms, the collected data is aggregated and stored in one standardized data warehouse. Once the manufacturing data is combined, it can be analyzed with a yield analytics software which also alerts yield and test engineers of any irregularities occurring in the data.
Many semiconductor manufacturers have several fabs, often in different countries, so another important question associated with siloed data emerges: Can the data from multiple semiconductor factories be combined?
Yes, if you use a standardized data warehouse structure at every site. Thus, users can connect to multiple data servers, accessing data form different locations in the same format with seamless integration, allowing correlation of data from various factories.
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Leave a Reply