ALBERT

Description: The hydrogen interaction kinetics of the GeP b1 defect, previously identified by electron spin resonance (ESR) as an interfacial Ge dangling bond (DB) defect occurring in densities ∼7 × 10 12  cm −2 at the SiGe/SiO 2 interfaces of condensation grown (100)Si/a-SiO 2 /Ge 0.75 Si 0.25 /a-SiO 2 structures, has been studied as function of temperature. This has been carried out, both in the isothermal and isochronal mode, through defect monitoring by capacitance-voltage measurements in conjunction with ESR probing, where it has previously been demonstrated the defects to operate as negative charge traps. The work entails a full interaction cycle study, comprised of analysis of both defect passivation (pictured as GeP b1 -H formation) in molecular hydrogen (∼1 atm) and reactivation (GeP b1 -H dissociation) in vacuum. It is found that both processes can be suitably described separately by the generalized simple thermal (GST) model, embodying a first order interaction kinetics description based on the basic chemical reactions GeP b1  + H 2  → GeP b1 H + H and GeP b1 H → GeP b1  + H, which are found to be characterized by the average activation energies E f  = 1.44 ± 0.04 eV and E d  = 2.23 ± 0.04 eV, and attendant, assumedly Gaussian, spreads σE f  = 0.20 ± 0.02 eV and σE d  = 0.15 ± 0.02 eV, respectively. The substantial spreads refer to enhanced interfacial disorder. Combination of the separately inferred kinetic parameters for passivation and dissociation results in the unified realistic GST description that incorporates the simultaneous competing action of passivation and dissociation, and which is found to excellently account for the full cycle data. For process times t a  ∼ 35 min, it is found that even for the optimum treatment temperature ∼380 °C, only ∼60% of the GeP b1 system can be electrically silenced, still far remote from device grade level. This ineffectiveness is concluded, for the major part, to be a direct consequence of the excessive spreads in the activation energies, ∼2–3 times larger than for the Si DB P b defects at the standard thermal (111)Si/SiO 2 interface which may be easily passivated to device grade levels, strengthened by the reduced difference between the average E f and E d values. Exploring the guidelines of the GST model indicates that passivation can be improved by decreasing T an and attendant enlarging of t a , however, at best still leaving ∼2% defects unpassivated even for unrealistically extended anneal times. The average dissociation energy E d  ∼ 2.23 eV, concluded as representing the GeP b1 -H bond strength, is found to be smaller than the SiP b -H one, characterized by E d  ∼ 2.83 eV. An energy deficiency is encountered regarding the energy sum rule inherent to the GST-model, the origin of which is substantiated to lie with a more complex nature of the forward passivation process than basically depicted in the GST model. The results are discussed within the context of theoretical considerations on the passivation of interfacial Ge DBs by hydrogen.