The question of the correlation between magnetization, band splittings, and magnetic circular dichroism (MCD) in the fundamental gap region of dilute magnetic semiconductors is examined experimentally and theoretically taking the case of wurtzite Ga(1-x)FexN as an example. Magnetization and polarization-resolved reflectivity measurements have been performed down to 2K and up to 7T for x = 0.2%. Optical transitions originating from all three free excitons A, B and C, specific to the wurtzite structure, have been observed and their evolution with the magnetic field determined. It is demonstrated that the magnitude of the exciton splittings evaluated from reflectivity-MCD data can be overestimated by more than a factor of 2, as compared to the values obtained by describing the polarization-resolved reflectivity spectra with appropriate dielectric functions. A series of model calculations shows that the quantitative inaccuracy of MCD originates from a substantial influence of the magnetization-dependent exchange interactions not only on the spin splittings of excitons but also upon their linewidth and oscillator strength. At the same time, a method is proposed that allows to evaluate the field and temperature dependencies of the magnetization from MCD spectra. The accurate values of the excitonic splittings and of the magnetization reported here substantiate the magnitudes of the apparent $sp-d$ exchange integrals in (Ga,Fe)N previously determined.
Download