Photoionization cross-section analysis for a deep trap contributing to current collapse in GaN field-effect transistors

Abstract

We perform a more incisive numerical analysis of the photon energy dependence of the photoionization cross section of a prominent deep trap (conventionally labeled as Trap 1), which had been found [Klein et al., Appl. Phys. Lett. 75, 4016 (1999)] to act as a contributor to current collapse phenomena in GaN metal-semiconductor field-effect transistors. The analytical expression for the photoionization cross section of Trap 1 is taken in the form of a convolution of a temperature-independent electronic part with a thermally broadened Franck–Condon (FC) factor, which also applies to the relevant regime of large lattice relaxation. For a direct comparison with earlier results, we specialize the present analysis to an electronic cross-section part represented by the Lucovsky model in combination with the semiclassical (Gaussian) approximation for the FC factor. In qualitative accordance with an earlier estimation by Klein et al. we obtain a value of EO≈1.9 eV for the classical optical ionization energy in combination with a full width at half maximum of 0.64 eV. The latter implies, on the assumption for the average phonon energy, to be of order 50 meV, an apparently unusually large magnitude, D≈1.1 eV, for the Franck–Condon shift. This parameter constellation is equivalent to a thermal ionization (electron binding) energy, ET=EO−D, of about ET≈0.8 eV. Such a location of Trap 1 near the middle of the upper half of the fundamental gap of GaN, ET≈Eg/4, is at clear variance to the earlier suggestion by Klein et al. for Trap 1 to be a midgap level. The present estimation offers a chance for detecting the Trap 1 also by deep-level transient spectroscopy measurements. An eventual availability of photoionization cross-section data for different temperatures is seen to be the prerequisite for a decisive reduction of residual uncertainties concerning the configuration coordinate diagram.