[Show abstract][Hide abstract] ABSTRACT: We develop an atomistic, nearest-neighbor sp3s* tight-binding Hamiltonian to
investigate the electronic structure of dilute bismide alloys of GaP and GaAs.
Using this model we calculate that the incorporation of dilute concentrations
of Bi in GaP introduces Bi-related defect states in the band gap, which
interact with the host matrix valence band edge via a Bi composition dependent
band anti-crossing (BAC) interaction. By extending this analysis to GaBiAs we
demonstrate that the observed strong variation of the band gap Eg and
spin-orbit-splitting (SO) energy with Bi composition can be well explained in
terms of a BAC interaction between the extended states of the GaAs valence band
edge and highly localized Bi-related defect states lying in the valence band,
with the change in Eg also having a significant contribution from a
conventional alloy reduction in the conduction band edge energy. Our calculated
values of Eg and SO are in good agreement with experiment throughout the
investigated composition range x less than 13%. In particular, our calculations
reproduce the experimentally observed crossover to an Eg < SO regime at
approximately 10.5% Bi composition in bulk GaBiAs. Recent x-ray spectroscopy
measurements have indicated the presence of Bi pairs and clusters even for Bi
compositions as low as 2%. We include a systematic study of different Bi
nearest-neighbor environments in the alloy to achieve a quantitative
understanding of the effect of Bi pairing and clustering on the GaBiAs
[Show abstract][Hide abstract] ABSTRACT: We use a nearest-neighbour sp<sup>3</sup>s* tight-binding Hamiltonian to investigate the electronic structure of GaAs-based dilute bismide and bismide-nitride alloys. We show that the observed strong variation of the band gap (E<sub>g</sub>) and spin-orbit splitting energy (Δ<sub>SO</sub>) with Bi composition in GaBi<sub>x</sub>As<sub>1-x</sub> is due primarily to a band-anticrossing interaction between the extended states of the host matrix valence band maximum and highly localised Bi-related defect states lying in the valence band, with the change in E<sub>g</sub> also having a significant contribution from a conventional alloy reduction in the conduction band edge energy. We calculate a crossover to an E<sub>g</sub> <; Δ<sub>SO</sub> regime at approximately 10.5% Bi composition in bulk GaBi<sub>x</sub>As<sub>1-x</sub>, in agreement with recent experimental studies of GaBi<sub>x</sub>As<sub>1-x</sub> epilayers grown on GaAs. Finally, we present calculations which show that the effects of N and of Bi are largely independent of each other in random GaBi<sub>x</sub>N<sub>y</sub>As<sub>1-x-y</sub> alloys, of relevance for future high efficiency photonic devices.
Transparent Optical Networks (ICTON), 2011 13th International Conference on; 07/2011
[Show abstract][Hide abstract] ABSTRACT: Photomodulated reflectance (PR) spectra of (B,Ga,In)As epilayers reveal unusual changes of the fundamental band gap PR line shape with temperature and hydrostatic pressure. We show that these changes arise because temperature variation or hydrostatic pressure shifts the conduction band edge (CBE) into resonance with boron-related cluster states. The resulting line shape changes are described by a level repulsion model which yields states of mixed character with an exchange of oscillator strengths. This model is corroborated by theoretical calculations which show a finite density of boron cluster states above the CBE at room temperature, with appropriate symmetry to couple to the CBE state.
[Show abstract][Hide abstract] ABSTRACT: We investigate the magnetotransport properties of n-type (B,Ga,In)As alloys and an n-type GaAs reference sample under hydrostatic pressure up to 16 kbar in the temperature range from 1.6 to 300 K. The free carrier concentration and the mobility of the reference sample are almost independent of the applied hydrostatic pressure. In contrast, the free carrier concentration and the mobility in B0.027Ga0.913In0.06As alloys drop by orders of magnitude over the accessible pressure range. The observations can be explained by assuming that a boron-related density of localized states exists in the vicinity of the conduction band edge of the alloy. The analysis of the pressure-induced changes of the free carrier concentration yields an image of the boron-related density of localized states with characteristic features that are found to be in good agreement with theoretical calculations using a linear combination of isolated states model. Our results provide strong evidence that boron-related states act as isovalent traps in (B,Ga,In)As alloys.
[Show abstract][Hide abstract] ABSTRACT: The replacement of As by N in GaAs introduces several pertubations, including both a change in potential at the N site and also long range strain relaxation in the crystal. It can be very useful to describe this perturbation in terms of the change in the Hamiltonian, ΔH, due to the introduction of N. Using plane-wave approaches, ΔH treats explicitly all atomic displacements in the structure. By contrast, a tight-binding approach allows a much simpler analysis. We illustrate the benefits of the tight-binding method by considering Ga(P,As:Bi) and GaNAs. We show that the tight-binding method provides a clear and quantitative explanation for many of the unusual electronic properties of these and related alloys, including the nonmonotonic variation of electron effective mass, me, and gyromagnetic ratio, ge, in GaNAs.
Journal of Physics Conference Series 08/2010; 242(1):012002.
[Show abstract][Hide abstract] ABSTRACT: The band-anticrossing (BAC) model has been widely applied to analyse the electronic structure of dilute nitride III-V-N alloys such as GaNxAs1−x. The BAC model describes the strong band gap bowing observed at low N composition in GaNxAs1−x in terms of an interaction between the GaAs host matrix conduction band edge and a higher lying band of localized N resonant states. In practice, replacing As by N introduces a range of N-related defect levels, associated with isolated N atoms, N–N pairs and larger clusters of N atoms. We show that the effect of such defect levels on the alloy conduction band structure is strongly dependent on the relative energy of the defect levels and the host conduction band edge. We first consider GaNxAs1−x, where we show that the unexpectedly large electron effective mass and gyromagnetic ratio, and their non-monotonic variation with x, are due to hybridization between the conduction band edge and specific nitrogen states close to the band edge. The N-related defect levels lie below the conduction band edge in GaNxP1−x. We must therefore explicitly treat the interaction between the higher lying GaP host Γ conduction band minimum and defect states associated with a random distribution of N atoms in order to obtain a good description of the lowest conduction states in disordered GaPN alloys. Turning to other alloys, N-related defect levels should generally lie well above the conduction band minimum in InNSb, with the band dispersion of InNSb then well described by a two-level BAC model. Both InP and InAs are intermediate between InSb and GaAs. By contrast, we calculate that N-related defect levels lie close to the conduction band minimum in GaNSb, and will therefore strongly perturb the lowest conduction states in this alloy. Overall, we conclude that the BAC model provides a good qualitative explanation of the electronic properties of dilute nitride alloys, but that it is in many cases necessary to include the details of the distribution of N-related defect levels to obtain a quantitative understanding of the conduction band structure in dilute nitride alloys.
Semiconductor Science and Technology 02/2009; 24(3):033001. · 1.92 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: III—N—As as well as III—N—P materials have been successfully employed in optoelectronic devices. Nitrogen impurities in the
dilute range have been investigated in the indirect gap semiconductor GaP since the 1960s, where they were used to produce
green light-emitting diodes. In recent years new developments in molecular beam epitaxy and metalorganic vapour phase epitaxy
growth methods have made it possible to introduce up to a few percent of nitrogen into direct gap GaAs and indirect gap GaP.
GaInNAs with N concentrations of about 1% has been used as the active material in vertical cavity surface emitting laser (VCSEL)
devices operating at telecommunication wavelengths. However, GaAs:N and GaP:N differ considerably in terms of their electronic
structure. In GaNAs the nitrogen impurity states in the doping regime lie resonantly in the conduction band. With increasing
N-content a strong redshift of the GaAs-like fundamental band gap (referred to as E
−) occurs. It is accompanied by the formation of an N-induced E
+ band, which blueshifts with increasing N. This repulsion behaviour of E
− and E
+ can be well parameterized by a simple two-level band-anticrossing model, which forms the basis of the 10 band k·p model successfully employed for describing the electronic states of III—N—As layers in laser structures in the vicinity of
the GaAs-like E
− band gap. The situation in GaP is somewhat different because the N levels in the doping regime are situated in the band gap
close to the X conduction band states (corresponding to the indirect gap), i.e. well below the Г conduction-band states of
GaP (corresponding to the direct gap). In other words, the order of the N states and the Г conduction band states is reversed.
With increasing N content, the lowest conduction band must then evolve from the N-like states in GaNP. The question then arises
as to whether the bandanticrossing model yields a good description of the lowest conduction band E
− in this case. If so, the lowest band gap will in a two-level band-anticrossing model acquire a large Г-like density of states
comparable to a direct semiconductor, which is a prerequisite for employing GaNP based heterostructures in the active region
of laser devices. In this review, we compare the electronic structure of GaNAs and GaNP, demonstrating that, in GaNP, the
Г character in the energy range of the N localized states is spread over a broad variety of transitions. This situation cannot
be properly parameterized by the simple band-anticrossing model and indicates that, in contrast to GaNAs, the lowest conduction
band states are not suitable to promote laser action in GaNP alloys.
[Show abstract][Hide abstract] ABSTRACT: We show using an sp3s* tight-binding model that the band anti-crossing (BAC) model describes well the evolution of the lowest N-related conduction states in ordered GaP1−xNx alloys, including the evolution of the Γ character with increasing x. We obtain a good description of the lowest conduction states in disordered GaPN structures by explicitly treating the interaction between the GaP host Γ conduction band minimum and defect states associated with a random distribution of N atoms. We find a very similar value for the total Γ character mixed into the N levels in the ordered and disordered cases, but a wider distribution of states with Γ character in the disordered case. We show that the band gap reduction with increasing composition is dominated by the increasing formation of N cluster states. Overall key features of the band structure can be well described using a modified BAC model which explicitly includes the broad distribution of N levels in disordered GaPN alloys.
[Show abstract][Hide abstract] ABSTRACT: The dependence of the optical band gap of InNxSb1-x and GaNxSb1-x on nitrogen content has been calculated using an sp(3)s(*) tight-binding Hamiltonian in large supercell calculations that explicitly include the effects of allowing a random distribution of nitrogen atoms. We calculate for InNSb that the energy levels associated with nitrogen complexes consisting of two or three atoms mostly lie above the InNSb conduction band edge (CBE). The steady reduction in the band gap of InNSb with increasing N composition is then well described using a two-level band-anticrossing model. The calculated effective mass of electrons in InNSb also smoothly decreases with composition, as predicted by the band-anticrossing model, due to the steady closure of the band gap and weak mixing with the N states. For GaNSb the situation is dramatic: the band gap and optical properties are shown to be strongly affected and highly sensitive to the distribution of the nitrogen atoms. We find that there is a wide distribution of N levels lying close to and below the CBE. The higher-lying N states push the CBE down in energy, as in GaAs, but the large number of lower-energy N states mix in strongly with the conduction band edge states, severely disrupting the band edge dispersion in GaNSb.
[Show abstract][Hide abstract] ABSTRACT: The dependence of the optical band gap of InNxSb1−x and GaNxSb1−x on nitrogen content has been calculated using an sp3s* tight-binding Hamiltonian in large supercell calculations that explicitly include the effects of allowing a random distribution of nitrogen atoms. We calculate for InNSb that the energy levels associated with nitrogen complexes consisting of two or three atoms mostly lie above the InNSb conduction band edge (CBE). The steady reduction in the band gap of InNSb with increasing N composition is then well described using a two-level band-anticrossing model. The calculated effective mass of electrons in InNSb also smoothly decreases with composition, as predicted by the band-anticrossing model, due to the steady closure of the band gap and weak mixing with the N states. For GaNSb the situation is dramatic: the band gap and optical properties are shown to be strongly affected and highly sensitive to the distribution of the nitrogen atoms. We find that there is a wide distribution of N levels lying close to and below the CBE. The higher-lying N states push the CBE down in energy, as in GaAs, but the large number of lower-energy N states mix in strongly with the conduction band edge states, severely disrupting the band edge dispersion in GaNSb.
[Show abstract][Hide abstract] ABSTRACT: In GaAs1-xNx, the band gap energy decreases very rapidly with x and the electron effective mass shows a quite unusual compositional dependence characterized by a sudden doubling for x congruent to 0.1%. In this work, we investigate the origin of this behavior by photoluminescence measurements under hydrostatic pressure in as-grown and hydrogenated GaAs0.9989N0.0011 samples. First, we show that two nitrogen pair states emitting at 1.488 and 1.508 eV contribute mainly, but to a different extent, in determining the steep increase in the electron mass observed for x congruent to 0.1%. Tight-binding supercell calculations assign the 1.488 eV levels to isolated N pairs and the 1.508 eV states to N pairs perturbed by a nearby N atom, in disagreement with previous attributions but consistent with the electron mass data. Second, photoluminescence at high hydrostatic pressure discloses that these N pair states show quite different rates of passivation by hydrogen. By combining these findings with the calculated lattice energies associated with each N complex, we conclude that strain relaxation is a key mechanism driving the interaction of hydrogen with N atoms in GaAs1-xNx.
Physical Review B 01/2008; 77:155213. · 3.66 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: We use a tight-binding Hamiltonian to investigate the variation of energy gap with nitrogen (N) composition in InSbN and GaSbN, including the effect on the energy gap due to a random configuration of N atoms. We find that the assumed distribution of N atoms does not significantly affect the calculated energy gap in InSbN. By contrast, the electronic properties of GaSbN are strongly dependent on the assumed N distribution, with N-related defect levels strongly perturbing the lowest conduction band states and energy gap.
[Show abstract][Hide abstract] ABSTRACT: The conduction band structure of BxGa1−xAs has a near-linear blueshift of the energy gap, which can be described using the virtual crystal approximation, but a dramatic increase in the band edge effective mass me* at low B composition, similar to that observed in GaNxAs1−x. We use a tight-binding model to show that isolated B atoms have little effect either on the band gap or lowest conduction band dispersion in BxGa1−xAs. In contrast, B pairs and clusters introduce defect levels close to the conduction band edge, which, through a weak band-anticrossing interaction, significantly reduce the band dispersion in and around the Γ point, thus accounting for the strong increase in me* and reduction in mobility observed in these alloys.
[Show abstract][Hide abstract] ABSTRACT: The effective gyromagnetic factor of electrons, g(e)(*), has been determined by Zeeman splitting measurements in a large number of GaAs1-xNx/GaAs (x < 0.7%) samples. Upon N incorporation, g(e)(*) shows first a sign reversal with respect to that of GaAs, then increases abruptly for a nitrogen concentration of order of 0.04%, and finally displays a nonmonotonic dependence on composition for higher x values. This behavior is well reproduced by a modified k center dot p model taking into account a nonmonotonic loss of Gamma character of the conduction band minimum due to multiple crossings between the redshifting conduction band edge and N-cluster states.
Physical Review B 12/2006; 74(24):245202. · 3.66 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: A tight-binding model of the electronic structure of substitutional nitrogen in GaAs, together with a variational description of quasilocalized nitrogen-induced electronic states near the conduction band edge, is used to calculate the nitrogen-related alloy scattering of conduction band electrons in the dilute nitride alloy, GaNxAs1−x. The electron mobility in the nondegenerate and degenerate doping regimes is calculated for bulk and quantum well geometries from the energy-dependent scattering rate using the Boltzmann transport equation in the relaxation-time approximation. Nitrogen cluster states are found to dominate the scattering near the conduction band edge and play a crucial role in limiting the electron mobility. In the experimentally relevant regime of degenerate doping and at nitrogen concentrations of 1 to 2%, the room-temperature mobility is found to be limited to values less than 300 cm2(V s)−1, in agreement with experimental measurements.