Clarifying roughness and atomic diffusion contributions to the interface broadening in exchange-biased NiFe/FeMn/NiFe heterostructures
Introduction
Exchange-biased NiFe/FeMn heterostructures are largely studied due to their high exchange bias field (Hex) and low coercive field (HC) values, which enhance their potential for technological applications, for example, in spin valve magnetic sensors. The exchange bias (EB) effect [1], [2], [3], [4], [5], [6], [7], [8], [9] is characterized by a shift in the hysteresis loop [M(H) curve] along the applied magnetic field axis. Discovered in 1956 [1], it generally emerges from an interfacial exchange coupling between ferromagnetic (FM) and antiferromagnetic (AF) materials, when this FM/AF composite sample is cooled down under an applied magnetic field (H) through the AF Néel temperature (TN) or when the film is deposited in a non-zero H. After almost six decades, several theoretical models have been proposed to explain EB; some focusing only on the interface phenomena [1], [2], [3], [10], [11], [12], [13], [14], [15], [16], whereas others have also considered volume contributions [17], [18].
Depending on the applied theory model, the Hex value can be influenced by the extent of the exchange interaction between the AF and FM spins at the interface, FM layer thickness, AF anisotropy, interface width, AF domain size and other factors, which in turn, depend on the roughness, crystalline texture, AF grain sizes, AF layer thickness, and others [19]. However, the dependence of Hex with respect to some of these factors is still unclear (in particular, the dependence on the roughness is not an exception) [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29]. For several systems, the Hex magnitude varies inversely proportional to the roughness value [23], [24], [25], [26], [27], while in others the opposite behavior is observed [28]. There are also systems that seem to be less sensitive to the roughness effect [29].
Specifically for the NiFe/FeMn system, we have previously investigated [26], [27] the dependence of Hex and HC on the σ (root-mean-square roughness + atomic grading at the interfaces) of the bottom (B) and top (T) NiFe/FeMn interfaces in the Si(100)/WTi(7 nm)/NiFe(30 nm)/FeMn(13 nm)/NiFe(10 nm)/WTi(7 nm) trilayers. Three samples were prepared at different working gas pressures (0.27, 0.67 and 1.33 Pa) in order to induce modifications on the σ-values. In Fig. 1, it is shown the hysteresis loops for samples prepared with 0.27 and 1.33 Pa. It was found that: (i) μ0Hex-values decreased from 4.1 to 2.9 mT (B-interface) and from 11.6 to 6.2 mT (T-interface); (ii) μ0HC increased from 0.3 to 0.6 mT (B-interface) and from 0.8 to 2.0 mT (T-interface), when σ of the B-and T-interfaces enhanced from 0.5 to 1.0 nm and from 0.9 to 2.7 nm, respectively, as the working gas pressure (PAr +) was augmented from 0.27 to 1.33 Pa [27].
It was also found [30] that the PAr + has a strong effect over the NiFe/FeMn interfaces, expressed as an enhancement of the σ-values. Based on thermodynamic principles, it was also assumed [30] that the main contribution to the σ-quantity comes from the atomic diffusion at the interfaces; however this assumption has not yet been proven. The separate contributions from roughness and atomic grading to the total interface disorder have not been quantified.
In order to elucidate the aspects mentioned above and, therefore, to completely discover the origin for the reduction of the Hex-values with PAr + in the NiFe/FeMn system, specular (SX) and off-specular longitudinal diffuse (OSX-diffuse) X-ray scattering and ferromagnetic resonance measurements were applied to the NiFe/FeMn bi- and trilayers to gain insight on the perpendicular correlation length and to distinguish the roughness and chemical grading effects at the interfaces; information that cannot be obtained when only SX-experiments are performed.
Section snippets
Experimental procedure
Si/WTi(7 nm)/NiFe(30 nm)/FeMn(13 nm)/WTi(7 nm) bilayers and Si/WTi(7 nm)/NiFe(30 nm)//FeMn(13 nm)/NiFe(10 nm)/WTi(7 nm) trilayers were prepared from Ni81Fe19, Fe50Mn50 and W90Ti10 targets by magnetron sputtering in the Thin Film Laboratory of LEMAG/UFES, using an ORION 8/AJA Sputtering System. The base pressure of the main chamber before the deposition was roughly at 6.7 × 10− 4 Pa. Three different PAr + were used: 0.27, 0.67 and 1.33 Pa. Ni81Fe19 and Fe50Mn50 layers were deposited by DC ignition, while W90Ti
Specular and diffuse scattering overview
One starts this section giving general information that will help readers to better understand our results, since comparisons among SX and OSX-diffuse experiments are not commonly discussed in the literature. In fact, we plan to show how SX and OSX-diffuse experiments can help us to distinguish contributions from the roughness and from the chemical grading effects at the interfaces. OSX-diffuse scattering has influences from interface roughness, but not from the atomic diffusion, while the
Results and discussions
Fig. 2 shows the SX and OSX-diffuse patterns for the B0.27-sample and their respective fittings as a function of the detector angle (α) related to the sample plane.
Two well-defined oscillations can be observed in both the SX and OSX-diffuse patterns: the first one with a low frequency, which can be associated with the scattering that comes from the WTi buffer and capping layers and the second one that is due to the multilayer total thickness (finite-size). The physical explanation for the
Conclusions
In this work, we have studied the physical origin for the reduction of the exchange bias field in sputtered NiFe/FeMn systems. Our goal was to separate the contributions from the roughness and from the atomic interdiffusion effect in both interfaces of the NiFe/FeMn/NiFe heterostructures prepared under three different argon working pressures (0.27, 0.67 and 1.33 Pa). We have combined specular and off-set longitudinal diffuse X-ray scatterings with ferromagnetic resonance data to understand the
Acknowledgments
This work was supported by the Brazilian agencies CNPq, FAPES and Brazilian Synchrotron Light Source (LNLS) under Proposal No. XRD2-6755. We acknowledge the LNLS staff for their prompt assistance at the XRD2 beamline.
References (48)
- et al.
J. Magn. Magn. Mater.
(1999) J. Magn. Magn. Mater.
(1999)- et al.
J. Magn. Magn. Mater.
(1984) - et al.
Appl. Surf. Sci.
(2007) - et al.
IEEE Trans. Magn.
(2005) - et al.
Appl. Phys. Lett.
(1996) - et al.
Physica B
(1998) Ultrathin Magnetic Structures II
(1994)- et al.
Appl. Surf. Sci.
(2012) - et al.
Phys. Rev. B
(2003)
Phys. Rev.
Phys. Rev.
J. Appl. Phys.
Phys. Rev. Lett.
J. Phys.
Ann. Phys. (Paris)
J. Appl. Phys.
Phys. Rev. B
J. Appl. Phys.
J. Appl. Phys.
Phys. Rev. Lett.
Phys. Rev. Lett.
J. Appl. Phys.
Europhys. Lett.
Appl. Phys. Lett.
Solid State Commun.
Phys. Rev. B
Phys. Rev. B
Phys. Rev. B
J. Appl. Phys.
J. Magn. Magn. Mater.
J. Appl. Phys.
J. Appl. Phys.
Cited by (3)
Characterization on surface and interface structure of epitaxial Pr<inf>0.7</inf>Sr<inf>0.3</inf>MnO<inf>3</inf>/La<inf>0.5</inf>Ca<inf>0.5</inf>MnO<inf>3</inf>/Pr<inf>0.7</inf>Sr<inf>0.3</inf>MnO<inf>3</inf> trilayer
2015, Superlattices and MicrostructuresCitation Excerpt :The performance of the advanced thin film devices often crucially depends on the microstructure of the multilayer films such as surface status, mass density and thickness of the component layer, interlayer roughness and growth orientation, etc. Specially, it should be pointed out that surface and interface status play an important role in FM/AFM magnetic coupling effect, which affects significantly the physical properties of manganite multilayer [3–6]. For example, electronic phase separation can be controlled by the presence of magnetic nanodots at the surface.
Effect of surface inhomogeneities on crystalline structure and magnetic properties of epitaxial Pr<inf>0.7</inf>Sr<inf>0.3</inf>MnO<inf>3</inf>/La<inf>0.5</inf>Ca<inf>0.5</inf>MnO<inf>3</inf> film on (001) SrTiO<inf>3</inf>
2015, Applied Physics A: Materials Science and Processing