Light is produced when a scanning tunneling microscope is used to probe a metal surface. Recent experiments on cobalt utilizing a tungsten tip found that the light is circularly polarized; the sense of circular polarization depends on the direction of the sample magnetization, and the degree of polarization is of order 10%. This raises the possibility of constructing a magnetic microscope with very good spatial resolution. We present a theory of this effect for iron and cobalt and find a degree of polarization of order 0.1%. This is in disagreement with the experiments on cobalt as well as previous theoretical work which found order of magnitude agreement with the experimental results. However, a recent experiment on iron showed $0.0\pm 2\%.$ We predict that the use of a silver tip would increase the degree of circular polarization for a range of photon energies.

• Received 28 April 1999

DOI:https://doi.org/10.1103/PhysRevB.61.3534