Realization of exchange bias control with manipulation of interfacial frustration in magnetic complex oxide heterostructures

Rich exchange bias (EB) behaviors were previously observed when ferromagnetic (FM) materials contacted a spin glass, demonstrating magnetic degrees of freedom of the coupling between the glass and FM spins. However, the correlation between the degree of magnetic spin frustration and the strength of the resulting EB is far from being understood. Here, we systematically investigate the dependency of EB on interfacial spin frustration in magnetic complex oxide heterostructures including ${\mathrm{La}}_{0.7}{\mathrm{Ca}}_{0.3}\mathrm{Mn}{\mathrm{O}}_3\text{/}\mathrm{CaMn}{\mathrm{O}}_3$ (LCMO/CMO) systems. The experimental analysis revealed that the extent of interfacial spin frustration is determined by the subtle competition between different types of magnetic orders related to the glassy spin behaviors at the interface. Such spin frustration can be manipulated through strain engineering through changes in the Mn $e_g$ orbital by alternating the stacking sequence of the heterostructures. A highly tunable EB field with 95% change of strength between the highly and weakly frustrated heterostructures has been achieved. Magnetic depth profiles of the heterostructures provide convincing evidence that a magnetically depressed region always occurs in the LCMO layer at the LCMO/CMO interfaces irrespective of the stacking sequence. Finally, EB is established at the magnetic interface in the LCMO layer.

Figure 1

Schematic diagrams of the interfacial spin alignments of exchange biased multilayers, depicting the collinearly aligned (a) ferromagnetic (FM)-coupled and (b) antiferromagnetic (AFM)-coupled interfaces, and the (c) noncollinearly aligned and (d) FM/spin-glass (SG) interfaces resulting an interfacial domain state and random pinning regions, respectively.

Figure 2

(a) X-ray diffraction, and (b) x-ray reflectivity data of the Type-C and Type-T heterostructures. Reciprocal space mapping (RSM) of the asymmetric lattice point around the (103) reflection of the ${\mathrm{SrTiO}}_3$ (STO) substrate is shown for the (c) Type-C and (d) Type-T heterostructures.

Figure 3

$M-H$ data (enlarged view) of the (a) Type-C and (b) Type-T heterostructures after zero-field cooling (ZFC; O Oe) and field cooling (FC; 150 Oe) procedures. $M-T$ data of the (c) Type-C and (d) Type-T heterostructures after ZFC and FC procedures in a range of measurement fields between 50 and 2000 Oe. The inset shows the corresponding $H^{2/3}$ vs $T_{\mathrm{irr}}$ plots, where the fitting (line) to the data (dot) is obtained using Eq. (1). Schematic representation of the ${\mathrm{MnO}}_6$ octahedral distortions influenced by (e) compressive strain and (f) tensile strain and subsequent energy degeneration of ${\mathrm{Mn}}^{3+} e_g$ subbands.

Figure 4

Magnetic relaxation data of the (a) Type-C and (b) Type-T heterostructures after field cooling (FC) measured at various temperatures. The measured data (dots) are fit (lines) using the standard stretched exponential function of Eq. (2).

Figure 5

The $H_{cf}$-dependent $M-H$ loops of the (a) Type-C and (b) Type-T heterostructures. The values of (c) $H_E$, (d) $H_C$, and (e) $M_E$ as a function of $H_{cf}$ are also plotted. Note that zero-field cooling (ZFC) results are displayed as $H_{cf}=1\mathrm{Oe}$ since the $x$ axis is plotted on a log scale. The (f) $H_E$, (g) $H_C$, and (h) $M_E$ as a function of temperature are also shown.

Figure 6

(a) A schematic of polarized neutron reflectometry (PNR) experimental setup. The top panels of (b) and (c) display the PNR reflectivity data of the Type-C and Type-T heterostructures, respectively. The experimental data are represented by the data points with error bars, and the best fits are represented by lines which were obtained through a least-squares fitting procedure. The bottom panels of (b) and (c) display the corresponding Neutron spin asymmetry data which is calculated using $(R^{++}-R^{--})/(R^{++}+R^{--})$.

Figure 7

The nuclear scattering length density (NSLD) and magnetic scattering length density (MSLD) model obtained from best fits to the polarized neutron reflectometry (PNR) data of the (a) Type-C and (b) Type-T heterostructures. The colored regions represent the different stacking sequences of the chemical layers within each heterostructure. The interfacial spin configurations of the heterostructures extracted from the analysis of the PNR data are depicted in (c) and (d) for the Type-C and Type-T heterostructures, respectively. A $M_d$ layer forms at the interface between the nonfrustrated ferromagnetic (FM) ${\mathrm{La}}_{0.7}{\mathrm{Ca}}_{0.3}{\mathrm{MnO}}_3$ (LCMO) layer and the antiferromagnetic (AFM) ${\mathrm{CaMnO}}_3$ (CMO) layer in each heterostructure. The depicted magnetic state of each heterostructures is one where the critical cooling field is applied, giving rise to a maximum $H_E$.