Abstract
The temperature dependence of the magnetization in fcc Fe on Cu(100) is calculated using a self-consistent local mean-field theory. The model reproduces an experimental magnetization oscillation as a function of film thickness and supports a picture where the top two layers are ferromagnetically coupled, and the remaining layers are antiferromagnetically coupled. The origin of the puzzling linear temperature dependence in oscillation amplitude is understood as a “surface phenomena” of the antiferromagnetic layer at the Fe/Cu interface. Proximity effects between a thin antiferromagnet with a low Néel temperature and a neighboring ferromagnet with a higher Curie temperature are discussed.