Magnetic exchange coupling in $\mathrm{IrMn}/\mathrm{NiFe}$ nanostructures: From the continuous film to dot arrays

A comprehensive description of the exchange bias phenomenon in an antiferromagnetic/ferromagnetic IrMn(10 nm)/NiFe(5 nm) continuous film and in arrays of square dots with different sizes (1000, 500, and 300 nm) is presented, which elucidates the temperature dependence of the exchange field $H_{\mathrm{ex}}$ and coercivity $H_C$, in conjunction with spatial confinement effects. To achieve this goal, samples prepared by electron beam lithography and lift-off using dc sputtering were subjected to structural investigations by electron microscopy techniques and to magnetic study, through SQUID and magneto-optic magnetometry measurements coupled to micromagnetic calculations. In particular, we have observed that at $T=300\mathrm{K}$ $H_{\mathrm{ex}}$ decreases by reducing the size of the dots and it is absent in the smallest ones, whereas the opposite trend is visible at $T=10\mathrm{K}$ ($H_{\mathrm{ex}}\sim 1140\mathrm{Oe}$ in the dots of $300\mathrm{nm}$). The exchange bias mechanism and its thermal evolution have been explained through an exhaustive phenomenological model, which joins spatial confinement effects with other crucial items concerning the pinning antiferromagnetic phase: the magnetothermal stability of the nanograins forming the IrMn layer (mean size $\sim 10\mathrm{nm}$), assumed as essentially noninteracting from the magnetic point of view; the proven existence of a structurally disordered IrMn region at the interface between the NiFe phase and the bulk of the IrMn layer, with a magnetic glassy nature; and the stabilization of a low-temperature $(T<100\mathrm{K})$ frozen collective regime of the IrMn interfacial spins, implying the appearance of a length of magnetic correlation among them.

Figure 1

Typical SEM image of the $\mathrm{IrMn}/\mathrm{NiFe}$ square dots. In particular, a portion of the array of dots with size of 300 nm is shown (the inset is an enlarged view).

Figure 2

(a) TEM image of the $\mathrm{IrMn}/\mathrm{NiFe}$ continuous film. The stack sequence is indicated. The short white lines underline the existence of a thin amorphous region (2–3 nm thick) with a light contrast, located between the IrMn crystalline phase and the amorphous NiFe layer. (b) High resolution image of the sample. The inset in the top-left corner is an enlarged view of a IrMn grain. The parallel white lines are used as in (a).

Figure 3

Exchange field $H_{\mathrm{ex}}$ (squares) and coercivity $H_C$ (circles) as functions of temperature T (5–300 K) measured by SQUID in the continuous $\mathrm{IrMn}/\mathrm{NiFe}$ film. In some cases, the error bar is smaller or comparable to the size of the dots. Solid lines are guides to the eye.

Figure 4

(a) Magnetization M vs T measured on the continuous $\mathrm{IrMn}/\mathrm{NiFe}$ film under a magnetic field $H_{\mathrm{inv}}=-50\mathrm{Oe}$ (full symbols); the thin line is M vs T for the reference NiFe sample, measured in a saturating magnetic field $H=50\mathrm{Oe}$. Both curves are normalized to the magnetization value at $T=5\mathrm{K}$, $M_0$. (b) Distribution of effective anisotropy energy barriers of the IrMn phase (normalized to the peak value) obtained by calculating the temperature derivative of the thick curve in (a), after normalizing to the values of $M/M_0$ of the thin line in (a). See text for explanation.

Figure 5

Exchange field $H_{\mathrm{ex}}$ (a) and coercivity $H_C$ (b) measured by MOKE as functions of T (10–300 K) on dots A, having $\mathrm{size}=1000\mathrm{nm}$ (circles), dots B with $\mathrm{size}=500\mathrm{nm}$ (squares), and dots C with $\mathrm{size}=300\mathrm{nm}$ (triangles). In some cases, the error bar is smaller or comparable to the size of the symbol. Solid lines are guides to the eye. (c) Hysteresis loops (normalized to the positive saturation value), measured by MOKE at $T=10\mathrm{K}$ on the array of dots A (circles), B (squares), and C (triangles).

Figure 6

Dependence of $H_{\mathrm{ex}}$ on the amount of AFM pinning centers as obtained by micromagnetic calculations simulating the magnetic behavior of a single squared dot with size 1000 nm (A type; circles) and 300 nm (C type; triangles).

Figure 7

Dependence of $H_{\mathrm{ex}}$ on the length λ of magnetic correlation among AFM pinning centers as obtained by micromagnetic calculations simulating the magnetic behavior of a squared dot with size $D=1200\mathrm{nm}$. In the inset $H_{\mathrm{ex}}$ is plotted as a function of the ratio $D/\lambda$.