The complexity of the silicon manufacturing processes has led to an explosion of data. Traditionally, engineering teams have had access to data pertaining to their step in the chip development process, but it’s been more challenging to obtain data from other phases of the chip’s lifecycle. More significantly, the raw data has been difficult to distill into useful insights. There’s a lot to sift through, and engineers need to know what to look for and what to query to make sense of it all. Considering the test data domain alone, there’s data stemming from wafer acceptance testing, bump, wafer sort, assembly, final test, and system-level test. There’s also critical importance in being able to tap into the data throughout the early design and manufacturing process, not just downstream. In short, both the depth and breadth of data support matters to help isolate and solve the root cause of any problems.
With semiconductor content rising in a number of application areas, there is growing urgency to move toward zero-defect approaches. The reality is, semiconductor defects are now commonly measured in parts per billion (ppb) rather than parts per million (ppm). Consider the automotive industry, where safety often hinges on the reliability and high performance of a vehicle’s electronic systems and semiconductor components. Even a seemingly miniscule defect rate can prove costly and potentially harmful and, hence, must be avoided. Never before has there ever been such an importance to accelerate convergence of quality and yield issues leveraging data analytics than now.