CPP magnetoresistance of magnetic multilayers: mean-free- path is not the culprit

In magnetoresistive (MR) studies of magnetic multilayers composed of combinations of ferromagnetic (F) and non-magnetic (N) metals, the magnetic moment (or related 'spin') of each conduction electron plays a crucial role, supplementary to that of its charge. While initial analyses of MR in such multilayers assumed that the direction of the spin of each electron stayed fixed as the electron transited the multilayer, we now know that this is true only in a certain limit. Generally, the spins 'flip' in a distance characteristic of the metal, its purity, and the temperature. They can also flip at F/N or N1/N2 interfaces. In this review we describe how to measure the lengths over which electron moments flip in pure metals and alloys, and the probability of spin-flipping at metallic interfaces. Spinflipping within metals is described by a spin-diffusion length, l M sf , where the metal M = F or N. Spin-diffusion lengths are the characteristic lengths in the current-perpendicular-to-plane (CPP) and lateral non-local (LNL) geometries that we focus upon in this review. In certain simple cases, l N sf sets the distance over which the CPP-MR and LNL-MR decrease as the N-layer thickness (CPP-MR) or N-film length (LNL) increases, and l F sf does the same for increase of the CPP-MR with increasing F-layer thickness. Spin-flipping at M1/M2 interfaces can be described by a parameter, δ M1/M2 , which determines the spin-flipping probability, P = 1 -exp(-δ). Increasing δ M1/M2 usually decreases the MR. We list measured values of these parameters and discuss the limitations on their determinations. OrganizationThis review is organized as follows. Section I provides a brief history and overview, defines the lengths of interest, briefly explains the physics underlying the spin-diffusion lengths that are the focus of the review, and discusses caveats on theoretical analysis and limitations on the measurements of the parameters of interest: the transport mean-free-path, λ t ; the spin-diffusion length in non-magnetic (N) metals or alloys, l N sf ; the spin-diffusion length in ferromagnetic (F) metals or alloys, l F sf ; and the interfacial spin-flip parameter, δ, where the spin-flipping probability is P = 1 -exp(-δ). Section II describes the different ways in which these parameters have been measured, and gives more specifics of their limitations. Section III contains four tables. Table I should be unique to each sample at 4.2K, but should be intrinsic in sufficiently high purity samples at 293K. Table III lists values of l F sf in ferromagnetic metals and alloys, mostly at 4.2K. Table IV lists values of δ N1/N2 for several N1/N2 metal pairs at 4.2K. Each table is preceded by some comments about the results. Section IV contains a brief summary and our conclusions. I. History, Overview, Definitions, and Caveats and Limitations. IA. History and Overview.The discovery in 1988 of Giant Magnetoresistance (GMR) in ferromagnetic/non-magnetic (F/N) metallic multilayers [1] [2] stimulated the growth of a new subfield of transport stud...

show abstract

“…(a) The data of [117] were confirmed in [16,52], so there is no issue about their correctness. The issue is solely their interpretation.…”

Section: Appendix C Consideration Of Mean-free-path Effects In Cpp Tmentioning

confidence: 91%

Spin-diffusion lengths in metals and alloys, and spin-flipping at metal/metal interfaces: an experimentalist’s critical review

Bass¹,

Pratt²

2007

J. Phys.: Condens. Matter

551

585

View full text Add to dashboard Cite

show abstract

“…It is reported that A⌬R of the multilayer, ͓FM1/NM/FM2/NM͔ N , is larger than that of ͓FM1/NM͔ N /͓FM2/NM͔ N , where FM1 and FM2 are ferromagnetic layers and NM is nonmagnetic layer. [12][13][14][15][16][17] However, A⌬R of our spin valves whose free and pinned layers are Fe 50 Co 50 inserted 1 nm Cu is as small as the simple Fe 50 Co 50 spin valve in series B. The spin memory loss cannot explain such a small A⌬R.…”

Section: Extended X-ray Absorption Structure Results and Discussionmentioning

confidence: 73%

Effect of inserted Cu on current-perpendicular-to-plane-giant magnetoresistance of Fe50Co50 spin valves

Yuasa

Fukuzawa

Iwasaki

et al. 2003

Journal of Applied Physics

View full text Add to dashboard Cite

We have reported that body-centered-cubic (bcc) Fe50Co50 with ultrathin Cu layers has large spin-dependent bulk scattering and enhances AΔR of spin valves. In order to clarify the role of Cu, two series of samples with (Fe50Co50 1 nm/Cu)×5 were prepared by two different sputtering machines. Machine A has a high rate, and machine B has a comparatively low rate. Both of AΔR in series A and B are increased by inserting thin Cu of about 0.1 nm, but they have a significant difference in Cu thickness dependence. While series A keeps high AΔR through the Cu thickness, AΔR in series B decreases when Cu is thickened. In order to find the reason for the difference, we took the extended x-ray absorption structure (EXAFS) spectra. By comparing AΔR and EXAFS, it was found that the very thin Cu of about 0.1 nm in Fe50Co50 takes a body-centered-cubic bcc structure, namely a solid solution state. Cu of 1 nm in spin valves with high AΔR takes a bcc and face-centered-cubic (fcc) evenly, and that with a low AΔR is a fcc structure. These results indicate that the existence of the solid solution Cu in Fe50Co50 is attributable to the enhancement of AΔR and are consistent with the report that Cu impurity in Fe and Co enhances the difference between the subband resistivity of up and down spins.

show abstract

“…The data for that study [37] are very similar to ours shown in figure 7 for identical Co/Cu structures. As illustrated in figure 8, we first showed that the inequality remained unchanged when the mean-free-path in the N-layer was reduced (by adding 2% Ge to Cu) from several times longer than the layer thicknesses to several times shorter [11]. We then showed that the inequality remained unchanged (or grew even greater) when the mean free paths in all three layers were reduced from larger to smaller than the layer thicknesses [38].…”

Section: Possible Spin-memory Loss At F/n Interfacesmentioning

confidence: 81%

“…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 t F and t N [7][8][9], 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.…”

Section: Introductionmentioning

confidence: 93%

Spin-Memory Loss in Metallic Multilayers

Pratt

Bass

2001

MRS Proc.

Self Cite

View full text Add to dashboard Cite

show abstract

CPP magnetoresistance of magnetic multilayers: mean-free- path is not the culprit

Cited by 22 publications

References 18 publications

Spin-diffusion lengths in metals and alloys, and spin-flipping at metal/metal interfaces: an experimentalist’s critical review

Spin-diffusion lengths in metals and alloys, and spin-flipping at metal/metal interfaces: an experimentalist’s critical review

Effect of inserted Cu on current-perpendicular-to-plane-giant magnetoresistance of Fe50Co50 spin valves

Spin-Memory Loss in Metallic Multilayers

Contact Info

Product

Resources

About