Static nontunneling models for the shallow donorsD(H,O) andD(Li,O) in germanium

Static models are proposed for the shallow donors D(H,O) and D(Li,O) in germanium as alternatives to the tunneling model used by Joós, Haller, and Falicov to interpret the photothermal ionization (PTI) spectra, and a critique of the tunneling model is made for these complexes. In the proposed model for D(H,O) a complex with trigonal symmetry that compresses the neighboring lattice along a threefold crystal axis yields a nondegenerate effective-mass (EM) ground state formed from the conduction-band valley parallel to the defect axis, and this state accounts for a sharp stress-independent PTI spectrum as observed in unstressed samples. As shown also by Broeckx, Clauws, and Vennik, a transition to a second sharp spectrum, as found experimentally, is predicted with this model for [111] uniaxial stress beyond a critical value determined by the electron deformation potential ${\Xi }_u$. A similar transition is predicted for [110] stress but not for uniaxial stress along [100]. For D(Li,O) a similar model in which the complex has only $C_{1h}$ symmetry and a single (110) plane of reflection symmetry is shown to account for the stress-independent splitting of the peaks in the high-stress spectrum previously attributed to tunneling by lithium. It is shown for both defects that the tunneling hypothesis imposes unlikely constraints on the model in order to fit the experimental spectra.