Effects of Multiple Slice Epitaxy and Front End Processing on Generation Threshold Current in Bubble Memory Devices

Abstract

The factors affecting generate threshold current, Ig, (the minimum current required to reliably nucleate a bubble under a hairpin generator) are reviewed. Ig data are presented for bubble memory devices fabricated using YSL-CaGe garnet (1) epitaxial films with nominal bubble sizes of 5.3 and 3.8 microns. Where multiple slice epitaxy employing large vertical stacks (up to 16 singly spaced slices/stack) is user it is lounct that Ig measured on an aluminum generator without overlying Permalloy elements or passivation is shown to depend on position in the epltaxy stack, decreasing by as much as 4 mA (1%) per slice position with increasing depth of slices in the melt. In contrast, slices ion milled to remove a 0.4 micron thick surface layer show no dependence of Ig on stack position. Removal of the surface layer has the additional effect of significantly decreasing the variation of Ig across a slice by up to 67%. These observations are interpreted in terms of the existence of a low anisotropy surface layer approximately 0.2-0.5 microns thick grown as slices are slowly withdrawn from the epi melt without rotation. The properties of this layer are discussed. In particular, the layer is shown to have a lower anisotropy than the bulk film, and it is estimated that the growth rate appropriate to the surface layer is approximately 20% of the value for normal film growth (0.4-1.0 microns/min). Finally the dependence of Ig on subsequent front end processing is presented, and it is shown that the Ig value measured on an aluminum generator without overlying oxide and Permalloy is reduced by approximately 30% after subsequent oxide and Permalloy processing steps.Generate threshold current (Ig), or the minimum current required to reliably nucleate a bubble under a hairpin generator, is a parameter that must be understood and controlled in order to produce reliable, interchangeable, and easily used bubble memory devices. Some of the factors contributing to Ig variations in devices have been discussed previously by Wolfe and Johnson (2) who have related Ig variation to stress effects in the epitaxial garnet layer due to the silicon oxide and the aluminum metallization and to changes in the stress because of the presence of the overlying layers, and through annealing effects arising during oxide and Permalloy deposition. This paper describes another major source of Ig variation that can be found in slices grown using multiple slice epitaxy employing large vertical stacks of slices, namely a contribution from an anomalous surface layer on the top of the epitaxial garnet film.