A Discussion of the Role of Distributed Effects in Latch-Up

Latch-up in monolithic integrated circuits is caused by the activation of four-layer SCR paths in certain technologies, in particular, bulk CMOS. A lumped element cross-coupled bipolar model has been widely used to discuss this parasitic SCR action with reasonable success. However, in the push toward VLSI structures, latch-up modeling and prediction based upon the simple model has proven inadequate. The deviations can be explained by an extension of the model into a network that couples parallel paths and by ensuring that measured parameters for the model reflect those of the latch-up situation. This paper discusses these points to allow the continued use of the lumped model in latch-up studies.