Spin-Memory Loss in Metallic Multilayers
- PDF / 167,103 Bytes
- 12 Pages / 612 x 792 pts (letter) Page_size
- 60 Downloads / 208 Views
Spin-Memory Loss in Metallic Multilayers W. P. Pratt, Jr., J.Y. Gu1, and J. Bass Department of Physics and Astronomy, Center for Sensor Materials, and Center for Fundamental Materials Research Michigan State University, East Lansing, MI 48824-1116 1 Present address: Materials Science Division, Argonne National Lab, Argonne, IL 60439-4845 ABSTRACT We review the results of measurements of spin-memory loss (spin-flipping) at cryogenic temperatures in sputtered or electrodeposited ferromagnetic and non-magnetic metals and alloys, and at metallic interfaces, using the current perpendicular to plane (CPP) geometry. INTRODUCTION Early models [1-5] of Giant Magnetoresistance (MR) in multilayers with alternating ferromagnetic (F) and non-magnetic (N) metals assumed that electrons can be divided into two classes: (1) those with magnetic moments in one direction (e.g. the direction of the first saturation magnetic field) and (2) those with moments in the opposite direction. It was also assumed that the moment directions stay fixed (moments do not flip) as the electrons traverse the multilayer. More quantitatively, these models assumed that the spin-flipping (spin-diffusion) lengths l sfF and l sfN [6] in the F- and N-metals are long compared to the layer thicknesses tF and tN, and that spin-flipping at the F/N interfaces can be neglected. Over the past several years, evidence from measurements of the current-perpendicular-toplane (CPP) MR has mounted that spin-diffusion lengths are not always longer than tF and tN [79], and that significant spin-flipping can occur at N1/N2 interfaces [9] and perhaps also at F/N interfaces [10, 11]. In this paper we describe that evidence and explain what we think we do and don’t yet know. We start with l sfF in F-metals and alloys, then turn to l sfN in N-metals and alloys, including superconductors, and conclude with spin-flipping at interfaces. We focus attention on spin-flipping at low temperatures (mostly at 4.2K), because little is yet known about how it varies with temperature [12], and low temperatures eliminate complicating contributions from magnons and phonons. Understanding spin-flipping is important for assessing technological potential, as it can limit the capabilities of multilayers for devices. Intriguingly, however, in certain cases [13] the presence of spin-flipping can enhance the change in specific resistance A∆R = ARAP – ARP, where A is the area of CPP current flow and RP and RAP are the sample resistances when the magnetizations of adjacent F-layers are aligned parallel (P) or anti-parallel (AP) to each other. With these definitions, CPP-MR = A∆R/ARP PRODUCING AN AP MAGNETIC STATE To compare CPP-MR data with models, one must reliably produce the P and AP magnetic states defined above. The P-state occurs naturally when the multilayer is taken to above the largest saturation field of the F-metals it contains. Achieving the AP state is more difficult. The authors of the first quantitative CPP-MR studies of simple Co/Cu, Co/Ag, and Py/Cu [8] F11.1.1
multilayers were apparent
Data Loading...
