Root Cause Searching on Yield Loss in the Automotive Industry by Multivariate Analysis and Modelling
In the context of the semiconductor manufacturing industry for automotive, the yield is monitored at wafer level as a KPI in terms of cost, but also in terms of quality as the link between quality and yield link is proven.
Here, a yield loss is observed at the electrical die-test step at hot temperature, and in particular, for a certain bin that is fitting with a group of tests of a specific component function. If a unit probing (UP) test, which the yield loss is observed for, highlights the failing dies, class probing (CP) tests the reticles, which are structures built between the dies and which monitor the manufacturing steps, before the dies were functional and may be tested. So searching for the root cause will target correlating the yield loss observed at UP with the CP tests to understand which manufacturing step is failing.
Using JMP Pro, correlation analysis, multivariate analysis, and modelling are implemented on UP and CP data for the failing lots; the results provided some good clues so that the device engineers could design corrective actions.
