Abstract
Measurements by Mössbauer spectroscopy of the position dependence of the hyperfine field in monocrystalline iron films show that the fractional deviation of the spontaneous magnetization at temperature T from its value at T=0 K is larger by a factor of about two at a film surface than in the film’s interior. This result agrees with an early theoretical prediction of a factor of exactly two which is based on the assumption that the surface anisotropy is zero. In contrast, the results of recent ferromagnetic resonance experiments on ultra-thin films of monocrystalline iron were shown to be dominated by a surface anistropy which is nonzero. This discrepancy is reconciled for measurements at T=300 K by making use of the general boundary condition which contains the exchange stiffness A and some component(s) of the surface anisotropy Ksurf. The crucial argument is that at 300 K the thermally excited spin wavelengths are so short that at the film surfaces the normal derivative 2A∂m↘/∂n of the oscillating magnetization m↘ is very much larger than Ksurfm↘. Thus Ksurfm↘ is neglible for thermal spin waves even though it is comparable to 2A∂m↘/∂n for the long decay distances (or wavelengths) occurring in ferromagnetic resonance.