Possibility of finding reliable solid-state tight-binding parameters for the Si-N bond through quantum-chemistry calculations

Abstract
Tight-binding parameters are obtained through a fit to the molecular levels and electronic populations of NH3, SiH4, Si(NH2 )4, and N(SiH3 )3 molecules. The resulting rms deviation of the adjusted molecular levels is only 0.31 eV, proving the goodness of the approach. The electronic structure of stoichiometric silicon nitride is calculated within a Bethe-lattice approach for our tight-binding parametrized Hamiltonian. Our results are discussed and compared with similar calculations using tight-binding parameter sets of several authors. Since our result is as good as or better than previously published electronic structures of a-Si3/7 N4/7, we proceed to further solid-state applications using an effective-medium approximation for the description of the electronic structure of silicon nitride alloys: (i) isolated N impurity in amorphous silicon in two different atomic configurations [planar (three bonds) and tetrahedral (four bonds)], (ii) the evolution of valence- and conduction-band edges of a-Si1x Nx as N content increases, (iii) the position of the Si dangling-bond defect level for the whole range of silicon nitride compositions (the N dangling bond does not introduce any defect level into the semiconducting gap), (iv) the electronic structure around Si-Si homopolar bonds embedded in stoichiometric silicon nitride, and (v) the electronic structure of isolated Si-H and N-H bonds and their behavior as a function of the [N]/[Si] ratio. At the end of the whole process, we reach the conclusion that although our tight-binding parametrization of the Si-N bond cannot be the final one, a large step towards internal consistency is given by our quantum-chemistry approach. We conclude by suggesting some careful experimental analysis that would eventually confirm the existence of the predicted state produced at Ev(a-Si)+0.55 eV by isolated N impurities in threefold coordination.