Influence of Si–N complexes on the electronic properties of GaAsN alloys

Abstract

Rights

Silicon dopant passivation by nitrogen during molecular beam epitaxy of GaNAs

“…Li et al [23] believed that it is Si Ga -N As that is formed during RTA based on a super cell simulation, while Janotti et al's [24] proposed a (Si-N) As split interstitial model which is more energetically favorable and reasonably accounts for the *0.8 eV photoluminescence (PL) peak observed on the annealed samples. On the other hand, early studies on GaAs:Si have suggested that both V/III ratio and growth temperature will affect Si doping efficiency based on the law of mass action [25]; however, they cannot account for the amount of efficiency decrease observed in our GaNAs samples as well as those in the literature [26,27]. Jin et al [26] also ruled out the possibility by comparing modulation-doped n-AlGaAs/GaAs and n-AlGaAs/GaNAs that the decrease in Si doping efficiency is due to N 2 dimer-related traps.…”

“…On the other hand, early studies on GaAs:Si have suggested that both V/III ratio and growth temperature will affect Si doping efficiency based on the law of mass action [25]; however, they cannot account for the amount of efficiency decrease observed in our GaNAs samples as well as those in the literature [26,27]. Jin et al [26] also ruled out the possibility by comparing modulation-doped n-AlGaAs/GaAs and n-AlGaAs/GaNAs that the decrease in Si doping efficiency is due to N 2 dimer-related traps. In this letter, we apply the mutual passivation model to the growth surface and calculated the doping efficiency by Monte Carlo simulation during molecular beam epitaxy (MBE).…”

“…When temperature is ramped up to inside the ''forbidden window,'' it enters the multilayer growth mode that ends up with a much rougher surface [36,37]. Therefore, it is reasonable to see the good agreement between Monte Carlo simulation and the experimental data below the ''forbidden window'' when N adatom motion can be described by lateral diffusion within a monolayer, while the simulation cannot account for mechanisms, such as islands formed in multilayer growth mode [37] and higher density of available group-V lattice sites on a (2 9 4) surface reconstruction at higher substrate temperature [compared to (2 9 1) at lower] [26] that can facilitate the bond forming in (Si-N) As besides the longer diffusion length levitated by growth temperature. Therefore, the Si doping efficiency turns out to be smaller than what the diffusion model predicts.…”

“…It is predicted that the enhanced electron effective mass should increase the density of states and benefit applications such as thermoelectrics; nevertheless, the benefit can be outweighed by the low doping efficiency of Si [9]. As Yu et al's [22] and Jin et al's [26] experiments and Janotti et al's [24] calculation suggest, group VI dopants are energetically unfavorable to form defect complex with N, and they are more practical to use when a high free electron concentration is desired.…”

Feasibility of enhancing the thermoelectric power factor in GaNxAs1−x

“…Therefore, bulk diffusion should not be sufficient. However, to explain the extensive passivation in our studies and that of others 33 it was necessary to speculate that surface diffusivity of Si adatoms is higher where ε ο is the permittivity of free space, ε r is the relative static permittivity of GaAs (= 13.1), n ~ (3-10)⋅10 17 /cm…”