Noncollinear and collinear magnetic structures in exchange coupled Fe/Cr(001) superlattices.

Abstract

The magnetic and structural properties of molecular beam epitaxy grown Fe/Cr(001) superlattices were studied as a function of the growth temperature ${\mathit{T}}_{\mathit{g}}$ using polarized neutron reflectometry (PNR) with polarization analysis, magneto-optic Kerr effect (MOKE), and x-ray-scattering techniques. From MOKE and PNR as a function of external field we find strong noncollinear coupling between the Fe layers and a so far unexpected coupling angle of 50\ifmmode^\circ\else\textdegree\fi{} near remanence for a sample grown at ${\mathit{T}}_{\mathit{g}}$=250 \ifmmode^\circ\else\textdegree\fi{}C. A detailed discussion of the domain structure of the sample near remanence confirms the modeling. On the other hand, an otherwise equivalent sample grown at room temperature exhibits completely ferromagnetic or uncoupled behavior. Using diffuse x-ray-scattering methods these distinct differences in the magnetic structure are found to be correlated with a growth temperature dependent length scale of constant Cr interlayer thickness ${\mathit{l}}_{\mathrm{Cr}}$. We find that ${\mathit{l}}_{\mathrm{Cr}}$ increases significantly with ${\mathit{T}}_{\mathit{g}}$. These results are discussed in the framework of current theories of noncollinear exchange. It is demonstrated that the bilinear-biquadratic formalism used so far is inconsistent with the data. The Cr specific proximity magnetism model is discussed which explains the occurrence of noncollinear coupling for systems with Cr interlayer thickness fluctuations on the length scale observed here for ${\mathit{T}}_{\mathit{g}}$=250 \ifmmode^\circ\else\textdegree\fi{}C. The model yields an exchange energy different from the bilinear-biquadratic formalism used so far, explaining the asymptotic approach to saturation observed by MOKE.