Is it the industry's intent to apply wafer-level burn-in (WLBI) to MPUs and ASICs ? Package-level burn-in (PLBI) today is facing escalating burn-in power dissipation for these MPU and ASIC devices. The burn-in board (BIB) density (devices / BIB) varies inversely with device power dissipation. However, wafer-level burn-in would indicate the opposite -- more devices in a 200 mm diameter. Is it the industry's intent to apply wafer-level burn-in to memory and more specifically, to DRAM ? Here,the power situation is much less problematic than MPUs and ASICs. Yet, unlike logic, DRAMs are overwhelmingly wirebonded die which require a test interface (probe) that can penetrate the bond pad's aluminum oxide without damaging the wafer. There are multi-device under test (DUT) probe technologies available today for x32 DUT testing; they can cover an area of four square inches. And it may be possible to leverage these technologies to provide full wafer contact -- but not for free. How would these compare to the PLBI economic model that we know and love ? In the 1993-94 timeframe, several technology issues concerning WLBI were considered risky and required years to solve. So where is the design point today? Is WLBI a static burn-in methodology where applying elevated voltage and temperature is sufficient ? Is it dynamic burn-in ? Monitored burn-in ? In-situ burn-in ? Or, will it naturally progress to wafer-level test and burn-in (WLTB) ? This paper provides an overview of wafer- level burn-in, past and present, and offers a prediction for the future.
