Coupling effects on current-assisted magnetization switching

Coupling effects on current-assisted magnetization switching

ARTICLE IN PRESS Journal of Magnetism and Magnetic Materials 272–276 (2004) 1586–1587 Coupling effects on current-assisted magnetization switching S...

179KB Sizes 2 Downloads 86 Views

ARTICLE IN PRESS

Journal of Magnetism and Magnetic Materials 272–276 (2004) 1586–1587

Coupling effects on current-assisted magnetization switching S. Urazhdin*, W.P. Pratt Jr, J. Bass Department of Physics and Astronomy, Center for Fundamental Materials Research, and Center for Sensor Materials, rm 4213 BPS bldg, Michigan State University, East Lansing, MI 48824-2320, USA

Abstract We show that antiferromagnetic or ferromagnetic coupling between Co layers strongly affects the switching behavior of Co/Cu/Co nanopillars, changing hysteretic switching at low magnetic field to nonhysteretic. We describe the behavior in terms of current-assisted thermal activation over a barrier. r 2003 Elsevier B.V. All rights reserved. PACS: 73.40.c; 75.60.Jk; 75.70.Cn Keywords: Current-driven switching; Magnetic temperature

Current-assisted switching in magnetic nanopillars has lately been receiving attention both for basic physics and technology (see Refs. [1,2], and references therein). Most studies are of uncoupled or antiferromagnetically (AF) coupled Co/Cu/Co nanopillars, and focus on similarities for these cases. We recently showed [3] that current-driven magnetization switching of uncoupled and AF-coupled nanopillars can be very different: hysteretic at low magnetic field H for uncoupled samples, but non-hysteretic (reversible) and characterized by telegraph noise for AF-coupled ones. Here, we show that the reversible switching at low H; characterized by telegraph noise, occurs both in AF coupled and ferromagnetically (F) coupled samples, and describe these behaviors in terms of current-assisted thermal activation. Our sample geometry and preparation procedures are described elsewhere [3]. For uncoupled samples, we use a Co(20)/Cu(10)/Co(2.5) trilayer, where thicknesses are in nm, and pattern the top Co(2.5) and most of the Cu(10) layer to nanopillar size (B70 nm  130 nm), leaving the bottom Co(20)-layer extended. Dipolar AF coupling was achieved by also patterning about 10 nm of the Co(20) layer. F coupling was achieved by reducing the Cu thickness to Cu(2.6), near the third RKKY *Corresponding author. Tel.: +1-517-355-9200X2233; fax: 1-517-353-4500. E-mail address: [email protected] (S. Urazhdin).

magnetoresistance (MR) minimum [4]. Sample shape variations and interfacial roughness led to variations in both AF and F coupling strengths. Differential resistances, dV =dI; were measured at Tph ¼ 295 K with four probes and lock-in detection [5]. Positive current flowed from the thick to the thin Co layer. Fig. 1 summarizes the differences between uncoupled (left), AF-coupled (middle), and F-coupled (right) samples. At I ¼ 0; the uncoupled (Fig. 1(a)) and AFcoupled (Fig. 1(d)) samples show the usual changes from a low resistance, high-H state in which the magnetizations of Co layers are aligned parallel (P), to a high resistance, low field state where they are aligned antiparallel (AP). In contrast, F-coupling causes the magnetizations to reverse simultaneously at small H; giving only a small feature in MR at I ¼ 0 (Fig. 1(h)). At large enough I > 0; F-coupled samples gave B5% MR, similar to the values for uncoupled or AF-coupled samples. Fig. 1(b), (e) and (i) compare the variations of dV =dI with I: At small H ¼ 20250 Oe, applied to fix the magnetization of the bottom Co layer, the uncoupled sample (Fig. 1(b), solid curve) shows hysteretic switching, while the AF-coupled (Fig. 1(e), top curve) and F-coupled (Fig. 1(i)) ones show reversible switching at Io0 and I > 0; respectively. At larger H; the switching in uncoupled samples (Fig. 1(b), dashed curve) becomes reversible, and in AF-coupled samples (Fig. 1(e), lower two curves) it first becomes hysteretic and then reversible again. Time resolved measurements of

0304-8853/$ - see front matter r 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.jmmm.2003.12.595

ARTICLE IN PRESS S. Urazhdin et al. / Journal of Magnetism and Magnetic Materials 272–276 (2004) 1586–1587

1587

I>0 AP

P (a)

P

AP

(b)

AP

(c)

P

Fig. 2. Schematics of current-driven switching over an effective magnetic barrier, as explained in the text. Dashed lines indicate the effective magnetic temperatures.

Fig. 1. Results for uncoupled (a–c), AF-coupled (d–g), and Fcoupled (h–j) samples. In the hysteretic plots, arrows show the scan direction. (a,d,h) dV =dI vs. H at I ¼ 0: (b,e,i) dV =dI vs. I at the listed values of H: In (e) curves are offset for clarity. (c,f,g,j) Time-resolved measurements of V =I at the listed values of I; H:

resistance at I; H near the reversible switching peaks show telegraph noise switching between AP and P states (Fig. 1(c), (f), (g) and (j)). At identical I of opposite signs, the average telegraph noise periods in the AF-coupled sample are similar. These different behaviors can be understood in terms of current-assisted thermal activation over an effective magnetic barrier U P;AP separating the AP and P states [3,5]. U P;AP are given by the magnetic anisotropy of the nanopillar, field H; and are also dependent on I: We assume that large enough current I > 0ðIo0Þ generates incoherent magnetic excitations in the P(AP) state, which become effectively thermalized due to the nonlinear magnetic interactions. The populations niPðAPÞ of excitations with energies Ei in the P(AP) states are then approximated by the Bose–Einstein distribution nPðAPÞ E½exp ðEi =kB TmPðAPÞ Þ þ 11 ; parameterized by i

TmPðAPÞ ; the effective magnetic temperatures in the P(AP) states. Fig. 2 illustrates the variations of U P;AP and TmP;AP for the different cases shown in Fig. 1, and the resulting thermally activated transitions. Fig. 2(a) is for hysteretic transitions. Both U P and U AP barriers are similar, but TmP > TmAP at I > 0; giving P-AP switching, and TmAP > TmP at Io0; giving AP-P switching. For the data acquisition time of 1 point/s, the thermally activated switching occurs at kB TmPðAPÞ E16U PðAPÞ [5]. With U P EU AP E1 eV [1], this gives Tm E700 K. For AF-coupling at low H (Fig. 2(b)), U P is small, U P o16 kB Tph ; and the P-AP switching is thermally activated at I ¼ 0: Large enough Io0 (causing TmAP > Tph ) activates the AP-P switching, leading to telegraph noise. Similarly, Fig. 2(c) shows that for F-coupling, or at large enough H in uncoupled and AF-coupled samples, the telegraph noise occurs at I > 0: We acknowledge helpful discussions with Norman O. Birge and support from the MSU CFMR, CSM, the MSU Keck Microfabrication facility, the NSF through Grants DMR 02-02476, 98-09688, and 00-98803, and Seagate Technology.

References [1] [2] [3] [4]

E.B. Myers, et al., Phys. Rev. Lett. 89 (2002) 196801. M.D. Stiles, A. Zangwill, Phys. Rev. B 66 (2002) 014407. S. Urazhdin, et al., Appl. Phys. Lett. 83 (2003) 114. S.S.P. Parkin, R. Bhadra, K.P. Roche, Phys. Rev. Lett. 66 (1991) 2152. [5] S. Urazhdin, et al., Phys. Rev. Lett. 91 (2003) 146803.