Antiferromagnetic domain wall control via surface spin flop in fully tunable synthetic antiferromagnets with perpendicular magnetic anisotropy

Antiferromagnetic (AF) domain walls have recently attracted revived attention, not only in the emerging field of AF spintronics, but also more specifically for offering fast domain wall velocities and dynamic excitations up to the terahertz frequency regime. Here, we introduce an approach to nucleate and stabilize an AF domain wall in a synthetic antiferromagnet (SAF). We present experimental and micromagnetic studies of the magnetization reversal in ${\left[{\left(\mathrm{Co}/\mathrm{Pt}\right)}_{X-1}/\mathrm{Co}/\mathrm{Ir}\right]}_{N-1}{\left(\mathrm{Co}/\mathrm{Pt}\right)}_X$ SAFs, where interface-induced perpendicular magnetic anisotropy (PMA) and AF interlayer exchange coupling (IEC) are completely controlled via the individual layer thicknesses within the multilayer stack. By combining strong PMA with even stronger AF-IEC, the SAF reveals a collective response to an external magnetic field applied normal to the surface, and we stabilize the characteristic surface spin-flop (SSF) state for an even number $N$ of AF-coupled ${\left(\mathrm{Co}/\mathrm{Pt}\right)}_{X-1}/\mathrm{Co}$ multilayer blocks. In the SSF state our system provides a well-controlled and fully tunable vertical AF domain wall, easy to integrate as no single-crystal substrates are required and with uniform two-dimensional magnetization in the film plane for further functionalization options, such as lateral patterning via lithography.

Figure 1

Schematic collective reversal modes in SAFs, where PMA energy $E_{K_u}$ and AF-IEC energy $E_{\mathrm{AF}-\mathrm{IEC}}$ dominate over dipolar interactions. For $E_{K_u}>E_{\mathrm{AF}-\mathrm{IEC}}$ the reversal proceeds via vertically and horizontally heterogeneous state of mixed FM/AFM stripe domains [3]. In contrast, for $E_{\mathrm{AF}-\mathrm{IEC}}>E_{K_u}$ the system shows heterogeneity during the reversal in only one direction normal to the film plane, while the system stays laterally homogeneous at all times (SSF state). Both reversal modes share the same laterally uniform AF layered remanent state. In our specific samples each arrow represents a FM ${\left(\mathrm{Co}/\mathrm{Pt}\right)}_3/\mathrm{Co}$ block and the coupling between adjacent blocks is mediated via 0.5-nm-thick Ir spacer layers.

Figure 2

Cross-sectional TEM images of the $N=20$ SAF film. (a) High-resolution STEM image of the Pt seed and the SAF, where all 80 Co layers are resolved. The white dashed lines indicate that columnar grains nucleate in the Pt seed and then subsequently proceed through the entire SAF structure. A true-to-scale sketch illustration of the complete SAF ML structure is shown at the right (blue: Pt; yellow: Ir; red: Co). (b) High-resolution TEM (HR-TEM) image, where the $[\mathrm{Co}\left(0.5\mathrm{nm}\right)/\mathrm{Pt}\left(0.7\mathrm{nm}\right){]}_3/\mathrm{Co}\left(0.5\mathrm{nm}\right)$ 4.1-nm-thick block structure separated by Ir(0.5 nm) layers becomes visible, as illustrated by the white labeling of the 20 FM blocks. (c) FFT of (a) with scale bar 1/nm. The white cones indicate the angular distribution of the correlated ML roughness and the crystallite alignment. The arrows mark the Co/Pt ML period reflection (white) and the Bragg peak positions of the out-of-plane lattice spacing for the Pt seed (red) and Co/Pt ML (blue), respectively.

Figure 3

Out-of-plane hysteresis loops measured by SQUID-VSM and calculated by micromagnetic mumax3 simulations for (a) $N=21$ and (b) $N=20$. The insets in (a) and (b) illustrate the corresponding magnetic configurations at the marked points (i)–(v) of model I. (c) shows the energy balance of anisotropy ($E_{K_u}$), exchange ($E_{\mathrm{exch}}$), and Zeeman ($E_{\mathrm{Zeeman}}$) energy, which add up to the total magnetic energy ($E_{\mathrm{tot}}$), for the refined model II and $N=20$. The exchange energy contains the exchange stiffness $A$ of the interatomic exchange as well as the IEC $J_{\mathrm{AF}-\mathrm{IEC}}$. The demagnetization energy is neglectable and set to zero here.

Figure 4

Comparison of out-of-plane hysteresis loops measured by SQUID-VSM and MOKE for (a) $N=21$ and (b) $N=20$. For convenience, the MOKE data are scaled to match the SQUID-VSM data in the high-field range.