On the limits of uniaxial magnetic anisotropy tuning by a ripple surface pattern

Abstract

Ion beam patterning of a nanoscale ripple surface has emerged as a versatile method of imprinting uniaxial magnetic anisotropy (UMA) on a desired in-plane direction in magnetic films. In the case of ripple patterned thick films, dipolar interactions around the top and/or bottom interfaces are generally assumed to drive this effect following Schlömann's calculations for demagnetizing fields of an ideally sinusoidal surface [E. Schlömann, J. Appl. Phys. 41, 1617 (1970)]. We have explored the validity of his predictions and the limits of ion beam sputtering to induce UMA in a ferromagnetic system where other relevant sources of magnetic anisotropy are neglected: ripple films not displaying any evidence of volume uniaxial anisotropy and where magnetocrystalline contributions average out in a fine grain polycrystal structure. To this purpose, the surface of 100 nm cobalt films grown on flat substrates has been irradiated at fixed ion energy, fixed ion fluency but different ion densities to make the ripple pattern at the top surface with wavelength Λ and selected, large amplitudes (ω) up to 20 nm so that stray dipolar fields are enhanced, while the residual film thickness t = 35–50 nm is sufficiently large to preserve the continuous morphology in most cases. The film-substrate interface has been studied with X-ray photoemission spectroscopy depth profiles and is found that there is a graded silicon-rich cobalt silicide, presumably formed during the film growth. This graded interface is of uncertain small thickness but the range of compositions clearly makes it a magnetically dead layer. On the other hand, the ripple surface rules both the magnetic coercivity and the uniaxial anisotropy as these are found to correlate with the pattern dimensions. Remarkably, the saturation fields in the hard axis of uniaxial continuous films are measured up to values as high as 0.80 kG and obey a linear dependence on the parameter ω2/Λ/t in quantitative agreement with Schlömann's prediction for a surface anisotropy entirely ruled by dipolar interaction. The limits of UMA tuning by a ripple pattern are discussed in terms of the surface local angle with respect to the mean surface and of the onset of ripple detachment.