Manipulation of competing ferromagnetic and antiferromagnetic domains in exchange-biased nanostructures

Using photoemission electron microscopy combined with x-ray magnetic circular dichroism we show that a progressive spatial confinement of a ferromagnet (FM), either through thickness variation or laterally via patterning, actively controls the domains of uncompensated spins in the antiferromagnet (AF) in exchange-biased systems. Direct observations of the spin structure in both sides of the FM/AF interface in a model system, $\mathrm{Ni}/\mathrm{Fe}{\mathrm{F}}_2$, show that the spin structure is determined by the balance between the competing FM and AF magnetic energies. Coexistence of exchange bias domains, with opposite directions, can be established in $\mathrm{Ni}/\mathrm{Fe}{\mathrm{F}}_2$ bilayers for Ni thicknesses below 10 nm. Patterning the $\mathrm{Ni}/\mathrm{Fe}{\mathrm{F}}_2$ heterostructures with antidots destabilizes the FM state, enhancing the formation of opposite exchange bias domains below a critical antidot separation of the order of a few $\mathrm{Fe}{\mathrm{F}}_2$ crystal domains. The results suggest that dimensional confinement of the FM may be used to manipulate the AF spin structure in spintronic devices and ultrahigh-density information storage media. The underlying mechanism of the uncompensated AF domain formation in $\mathrm{Ni}/\mathrm{Fe}{\mathrm{F}}_2$ may be generic to other magnetic systems with complex noncollinear FM/AF spin structures.

Figure 1

XMCD images of the domain structure of the bare 70-nm-thick $\mathrm{Fe}{\mathrm{F}}_2$ layer measured at the Fe edge (top left), and of Ni (3, 5, and 7 nm)/$\mathrm{Fe}{\mathrm{F}}_2$ (70 nm) bilayers measured at the Ni edge, at zero magnetic field and 30 K upon ZFC from 296 K. The Fe image shows the typical domain structure of the uncompensated Fe moments at the surface of $\mathrm{Fe}{\mathrm{F}}_2$. On the Ni images, the initial remnant saturated state (dark contrast) splits into a branched pattern of small, inverted domains with opposite magnetic orientation (bright contrast), as indicated by the arrows. The scale bar is the same for all images.

Figure 2

Temperature-dependent XMCD images from the same areas at the Ni and Fe edges of a Ni(4 nm)/$\mathrm{Fe}{\mathrm{F}}_2$(70 nm) sample, measured at zero magnetic field above and below $T_{\mathrm{N}}$. The measurements at 30 K, recorded upon ZFC from a remnant saturated state at 296 K, show how the domain structure of the uncompensated Fe moments (b) clearly replicates the inverted Ni domain pattern (a). On warming up above $T_{\mathrm{N}}$, both the inverted Ni domains (c) and those of the uncompensated Fe spins in the AF (d) are erased. The red arrows in the images point at a defect that enables one to identify the same area in the sample. The scale bar is the same for all images.

Figure 3

Temperature dependence of the magnetization for different cooling fields of the $\mathrm{Ni}(6\mathrm{nm})/\mathrm{Fe}{\mathrm{F}}_2$(70 nm) bilayers, showing a clear drop of the Ni magnetization below the $T_{\mathrm{N}}$ of $\mathrm{Fe}{\mathrm{F}}_2$. The drop scales with the strength of the cooling field, and the complete reversal of the Ni magnetization is reached for 1 kOe FC where only PEB is present (see also Fig. S6, Supplemental Material). The insets show XMCD images of the Ni domain structure close to the AF transition point (70 K), a few degrees below (65 K), and well below (30 K), measured at zero magnetic field upon ZFC from 296 K. The scale bar ($5\mu \mathrm{m}$) is the same for the three images.

Figure 4

Fraction of inverted domains for $\mathrm{Ni}/\mathrm{Fe}{\mathrm{F}}_2$(70 nm) bilayers as a function of Ni thickness, measured at zero field and 30 K, after ZFC from 296 K. The error bars represent the standard deviation of the mean but do not include smaller systematic errors arising from either the PEEM measurements (e.g., due to an inhomogeneous illumination of the XMCD images) or the analysis procedure (caused by image drift correction and evaluation of the area of the domains).

Figure 5

XMCD images at the Ni edge of the domain structure of square arrays of 200 nm in length square antidots with antidot densities $\mathrm{AD}=12$ and 24% patterned into a $\mathrm{Ni}(6\mathrm{nm})/\mathrm{Fe}{\mathrm{F}}_2\left(70\mathrm{nm}\right)$ sample, measured at zero field and 30 K after ZFC from 296 K. The arrows indicate the magnetization direction. Note that the gray contrast represents the absence of XMCD signal at the position of the antidots. The scale bar is the same for both images.

Figure 6

Fraction of inverted domains for $\mathrm{Ni}(6\mathrm{nm})/\mathrm{Fe}{\mathrm{F}}_2\left(70\mathrm{nm}\right)$ as a function of the AD, measured at zero field and 30 K, upon ZFC from 296 K (red) and FC (black; $H_{\mathrm{FC}}=50\mathrm{Oe}$ from 200 K to 45 K and $H_{\mathrm{FC}}=0\mathrm{Oe}$ from 45 K to 30 K). The error bars represent the standard deviation over up to 15 independent measurements. The occasional data points where no error bar is given refer as to cases where a comparable statistics is not available. The dashed lines are guides to the eye.

Figure 7

Correlation length of the inverted domains for $\mathrm{Ni}(6\mathrm{nm})/\mathrm{Fe}{\mathrm{F}}_2\left(70\mathrm{nm}\right)$ as a function of the AD, measured at zero field and 30 K, upon ZFC from 296 K (red) and FC (black; same protocol as in Fig. 6). The dashed line is a guide to the eye.