Is burn-in burned out?

Summary form only given, as follows. Burn-in screening was introduced almost simultaneously with the introduction of the monolithic integrated circuit in the early 1960´s. At the time, high incidences of surface contamination, predominantly in the form of mobile ions, bulk defects in active regions, and poor lead bonds made burn-in and related screens almost mandatory to achieve useful failure rates. Indeed, the magnitude of the problem is reflected by the 10% (100,000 ppm) Percent Defective Allowable (PDA) which was frequently specified. In the intervening years, orders of magnitude of improvement have been achieved in contamination levels, bulk defect densities, and interconnection integrity. Mature State-of-Art microcircuit manufacturing facilities using up to date process control methodology produce parts with bum-in failure rates against the failure mechanisms of concern in application of less than 100 ppm. Analysis of parts which have failed during the burn-in process indicates a high level (several thousand ppm) of induced failures associated with handling and testing and suggests that burning in low defect populations leads to the stocking and use of parts with lower quality and reliability than was originally present. Clearly, a screen which results in the creation of a higher failure rate in the application, not to mention the cost of performing the screen and the cost of the good parts which are destroyed, is not desirable.