Over-acceleration of corrosion mechanisms during reliability testing: A method to relate biased HAST tests and application conditions for Cu wire products

Cu bond wires in microelectronics have great potential but also provides several challenges. The acceleration factors or failure mechanisms in reliability tests are somewhat different with respect to gold wire bonding which, using unchanged but not validated test conditions and duration requirements, may lead to non-justified failures. With copper wire technology, the intermetallic compounds (IMC's) that form between the bond ball and the bond pad change in composition and corrosion behavior when compared to the gold wirebonding IMC's. When exposed to high temperatures, high moisture levels and high bias, these three different stress factors can add up to very high acceleration factors. When a product or material system fails in this test the question arises what the acceleration factor actually is and how this test compares to application conditions where temperatures may also rise significantly while high humidity levels may still be present and bias is applied. This could the operational startup phase after a long time of being in off or standby stage. The product will suddenly heat up due to the internal heat generation but moisture might still be present in the package. Combined with a high bias voltage this could lead to conditions as seen in the HAST test. To make a comparison between application and test, data was collected on the moisture properties of several molding compounds as a function of temperature. This data was then used in combination with thermal transient simulations of a product in application to compare actual moisture levels under use conditions to moisture levels in test. The simulation shows that the HAST test condition never occurs in the actual application and that the test condition is unrealistically accelerating due to the very high moisture loading. Less extreme conditions will be proposed and discussed. Finally some actual corrosion data will be shown that proof the validity of the simulation results.