Formation and dissolution of D-N complexes in dilute nitrides

Deuterium (hydrogen) incorporation in dilute nitrides (e.g., GaAsN and GaPN) modifies dramatically the crystal's electronic and structural properties and represents a prominent example of defect engineering in semiconductors. However, the microscopic origin of D-related effects is still an experimentally unresolved issue. In this paper, we used nuclear reaction analyses and/or channeling, high resolution x-ray diffraction, photoluminescence, and x-ray absorption fine structure measurements to determine how the stoichiometric [D]∕[N] ratio and the local structure of the N-D complexes parallel the evolution of the GaAsN electronic and strain properties upon irradiation and controlled removal of D. The experimental results provide the following picture: (i) Upon deuteration, nitrogen-deuterium complexes form with [D]∕[N]=3, leading to a neutralization of the N electronic effects in GaAs and to a strain reversal (from tensile to compressive) of the N-containing layer. (ii) A moderate annealing at 250 °C gives [D]∕[N]=2 and removes the compressive strain, therefore the lattice parameter approaches that of the N-free alloy, whereas the N-induced electronic properties are still passivated. (iii) Finally, annealings at higher temperature (330 °C) dissolve the deuterium-nitrogen complexes, and consequently the electronic properties and the tensile strain of the as-grown GaAsN lattice are recovered. Therefore, we conclude that the complex responsible for N passivation contains two deuterium atoms per nitrogen atom, while strain reversal in deuterated GaAsN is due to a complex with a third, less tightly bound deuterium atom.

Dilute nitride based double-barrier quantum-well infrared photodetector
operating in the near infrared

Near-infrared detection is reported for a double-barrier quantum-well infrared photodetector based on a 30-Å GaAs1−yNy (y ≈ 0.01) quantum well. The growth procedure using plasma-assisted molecular-beam epitaxy is described. The as-grown sample exhibits a detection wavelength of 1.64 μm at 25 K. The detection peak strengthens and redshifts to 1.67 μm following rapid thermal annealing at 850 °C for 30 s. The detection peak position is consistent with the calculated band structure based on the band-anticrossing model for nitrogen incorporation into GaAs

Terahertz response of hot electrons in dilute nitride Ga(AsN) alloys

We model theoretically an unusual ac negative differential mobility (NDM) effect that occurs when electrons are accelerated by an electric field in the highly nonparabolic conduction band of dilute nitride Ga(AsN) alloys. By solving balance equations that take into account the negative effective mass of electrons and the velocity and energy relaxation processes, we derive an expression for the maximum response frequency, fmax, associated with the NDM. Our predicted values of fmax depend on material composition and can be tuned by the applied electric field up to terahertz frequencies

Modeling of band gap properties of GaInNP alloys lattice matched to GaAs

Compositional and temperature dependences of the band gap energies of GaInNP alloys, which are lattice matched to GaAs, are determined and modeled by a band anticrossing (BAC) interaction between the localized state of the isolated NP and extended host states. The BAC parameters are deduced as EN = 2.1±0.1 eV and CMN = 1.7±0.2 eV. The low value of the coupling parameter CMN implies weaker coupling of the N level with the host matrix, presumably due to short range ordering effects, similar to the case of GaInNAs alloys with a high In content. The obtained information is important for future modeling of the electronic structure of the alloys

ROSSANA HERNANDEZ

electronica del estado solido

http://apl.aip.org/applab/v88/i3/p031907_s1?isAuthorized=no

---

Overannealing effects in GaInNAs(Sb) alloys and their importance to laser applications

The photoluminescence efficiency and linewidth are well-established metrics for characterizing potential laser active regions. We demonstrate the critical importance of a new parameter for predicting the performance of dilute-nitride lasers: the "optimal" postgrowth annealing temperature, defined as the annealing temperature giving the highest photoluminescence efficiency. We validate this assertion with two 1.55 μm edge-emitting GaInNAsSb lasers containing active regions with different optimal annealing temperatures. Although both active regions showed comparable photoluminescence efficiency and linewidth under optimal annealing conditions, laser performance was significantly different. The room-temperature threshold current density for the active region with higher optimal annealing temperature was 630 A/cm2, compared with 2380 A/cm2 for the sample with lower optimal annealing temperature. We conclude that overannealing of the gain region during upper cladding growth is the responsible mechanism. The dependence of the optimal annealing temperature on composition and growth conditions is also discussed.

ROSSANA HERNANDEZ
Electronica del estado solido
http://apl.aip.org/applab/v88/i22/p221115_s1?isAuthorized=no

---

Nitrogen-hydrogen complex in GaAsxN1−x revealed by x-ray absorption spectroscopy

GaAsN alloys belong to a class of semiconductors with fascinating physical properties. Indeed, a small amount of nitrogen incorporation in GaAs leads to a counterintuitive and large band-gap reduction, and to an unexpected sudden increase in the effective mass of electrons. Even more surprisingly, both electronic and structural changes can be reversed fully and in a tunable manner by hydrogen incorporation. In this paper, we combine x-ray absorption spectroscopy at the nitrogen edge with ab initio simulations to investigate the atomic geometry of N-H complexes in hydrogenated GaAsN. In this way, we provide experimental evidence that dihydrogen-nitrogen complexes with C2v symmetry are the most abundant species in hydrogenated GaAsN. This finding contradicts previous predictions of "in-line" N-H2* complexes as the predominant species, and accounts for recent infrared absorption experiments.

A dilute nitride compound for midinfrared optoelectronic devices

The growth of the antimony-rich dilute nitride alloys GaSbN, InGaSbN, and InGaAsSbN on GaSb substrates by solid-source molecular-beam epitaxy is reported. Bulk GaSbN layers are characterized by x-ray diffraction and the nitrogen incorporation is estimated to be close to 1%. A nitrogen-induced redshift of the photoluminescence peak wavelength as large as 110 meV is observed in In.15Ga.85Sb1−xNx/GaSb quantum wells. Photoluminescence emission at 77 K for an In0.3Ga0.7As0.1Sb0.9−xNx/Al0.25Ga0.75As0.02Sb0.98 multiple quantum well structure shows a 66 meV redshift due to nitrogen incorporation (2.33 μm emission wavelength), demonstrating the potential of this compound for midinfrared optoelectronic device applications

Growth and applications of Group III-nitrides

Recent research results pertaining to InN, GaN and AlN are reviewed, focusing on the different growth techniques of Group III-nitride crystals and epitaxial films, heterostructures and devices. The chemical and thermal stability of epitaxial nitride films is discussed in relation to the problems of deposition processes and the advantages for applications in high-power and high-temperature devices. The development of growth methods like metalorganic chemical vapour deposition and plasma-induced molecular beam epitaxy has resulted in remarkable improvements in the structural, optical and electrical properties. New developments in precursor chemistry, plasma-based nitrogen sources, substrates, the growth of nucleation layers and selective growth are covered. Deposition conditions and methods used to grow alloys for optical bandgap and lattice engineering are introduced. The review is concluded with a description of recent Group III-nitride semiconductor devices such as bright blue and white light-emitting diodes, the first blue-emitting laser, high-power transistors, and a discussion of further applications in surface acoustic wave devices and sensors.

ROSSANA HERNANDEZ

electronica del estado solido

http://iopscience.iop.org/0022-3727/31/20/001

---

Ga(N,As)/GaAs QWs

An interesting implication of the band-structure modification due to N in Ga(N,As) and (Ga,In (N,As) is the strong nonparabolicity of the conduction band predicted by theor.In comparison with GaAs, the electron effective mass is expected to increase by about 50% at k=0 for x≈1%, with a further mass increase for k≠0. There are, to date, several experimental studies of this interesting effect. Skierbiszewski et al. [51] observed a strong increase of the electron effective mass at the Fermi level in (Ga,In)(N,As):Se with increasing free-electron concentration up to a value of 0.4 m0 for n=6×1019 cm−3. Hai et al. [52,53] showed by cyclotron resonance on 7-nm GaNxAs1−x/GaAs QWs that the electron effective mass at the conduction band edge in GaNxAs1−x increases with x, whereas the hole effective masses are similar to those of GaAs. They reported values of 0.12 m0 and 0.19 m0 for x=1.2% and 2.0%, respectively. Wu et al. [54] pointed out that the confined states of Ga(N,As)/GaAs QWs can only be correctly described when the Ninduced changes of the electron effective mass are taken into account. Baldassari Höger von Högersthal et al. [55] reported a value of 0.15 m0 for x=1.6% determined by magnetophotoluminescence. A sharp increase of the electron effective mass was observed for N contents below 0.5% [56]. Similar experiments were performed by Wang et al. [58]. The electron effective mass is usually reported to strongly increase with increasing N content, apart from a report of transport experiments where the opposite trend was claimed [57]. The experimental data scatter considerably. This is partly a result of different assumptions underlying the models used for extracting the actual effectivemass values from the experimental data obtained by different experimental techniques. In addition, as discussed below in detail, effective-mass results obtained on bulk samples and quantum wells

of the same x are not necessarily comparable. The effective-mass issue is discussed from a different perspective.

A further complication arises because the conduction band of GaNxAs1−x is strongly nonparabolic. This can also be understood qualitatively in the framework of the levelrepulsion model. The closer the conduction-band states of the host are to the N level, the stronger is the level repulsion. Consequently, this means for QWs that the effective masses of the electron subbands in the conduction band must depend on N content, well width, and confinement energy. This manifests itself in the hydrostatic-pressure dependence of the interband transitions enhhn of Ga(N,As)/GaAs QWs, as can be seen in Figure 5.7. The figure shows series of PR spectra obtained under hydrostatic pressure at 300 K for two GaN0.018As0.982/GaAs QWs of width 8 nm and 4 nm. In the first series of spectra in Figure 5.7a, three signals can be clearly detected at all pressures; a fourth one can be discerned in the spectra for pressures exceeding 0.7 GPa. The signal at the highest energy originates from the GaAs barrier.

ROSSANA HERNANDEZ

Electronica del estado solido

http://www.torrent2crazy.com/search/physics+and+applications+of+dilute+nitrides

---

LOCAL DISORDER EFFECTS IN GaInNAs

A further source of disorder must be considered when analyzing the quaternary alloy, Ga1−yInyNxAs1−x. The nature of the isolated resonant state changes with local environment, depending on the average number of In and Ga atoms neighboring each N atom [42,43]. To investigate this variation, we took a number of 216-atom supercell structures, in which we constrained the central group V site to have a given number, m, ofindium nearest neighbors (m=0 to 4) [43], and then placed In atoms at random on the remaining sites to give an overall indium fraction of y=0.25.

We found, for the structures considered, that EN varies approximately linearly from EN=1.52 eV for four Ga neighbors to EN=1.75 eV for four In neighbors. The value of EN depends weakly on the atomic arrangement in the second shell of group III neighbors, but the calculated effect for a range of supercell calculations indicates that the atomic arrangement in the second shell only shifts EN by ~±0.021 eV, down by an order of magnitude on the effects of variations in the nearest-neighbor environment. The matrix element linking the N resonant state and the conduction band edge is calculated in the tight-binding method to vary between 2.00 x1/2 eV (for a structure where all N have four Ga neighbors) to 1.35 x1/2 eV (for a structure where all N have four In neighbors), with the conduction-band-edge energy, Ec, in the 2×2 matrix of Equation 3.1 then varying between 1.10 eV and 1.11 eV for the 216-atom supercells considered, and the band-gap between 0.95 eV and 1.04 eV as the number of In nearest neighbors m increases from 0 to 4. The average interaction in a larger unit cell containing a mixture of nearest-neighbor environments is then given by an appropriately weighted average of the above parameters. This justifies the construction of an effective two-band model to describe the variation of conduction-band-edge energy and energy gap in quaternary GaInNxAs1−x alloys, but the magnitude of the parameter β in the interaction term βx1/2 and the value of EN will vary, depending on the average N cation nearest-neighbor environment [25]. The experimentally observed blueshift in the energy gap of many GaInNAs samples with mild thermal annealing [25,39–41] can then be explained by local atomic rearrangements leading to an increase in the average number of In neighbors about each N atom [59–61].

ROSSANA HERNANDEZ

Electrónica del estado solido

http://apl.aip.org/sso/servlet/primaryLogin?url=http://apl.aip.org/applab/v87/i18/p182114_s1

---

Intrinsic limits on electron mobility in dilute nitride semiconductors

A fundamental connection is established between the composition-dependence of the conduction band edge energy and the n-type carrier scattering cross section in the ultradilute limit for semiconductor alloys, imposing general limits on the carrier mobility in such alloys. From the measured nitrogen composition dependence of the bandgap inGaAs_1−xN_x, the carrier scattering cross section of substitutional nitrogen defects in GaAs is estimated to be 0.3 nm². Within an independent scattering approximation, the carrier mobility is then estimated to be ∼ 1000 cm²/V s for a nitrogen atomic concentration of 1%, comparable to the highest measured mobility in high-quality GaInNAs samples at these N concentrations, but substantially higher than that found in many samples. This gives an intrinsic upper bound on the carrier mobility in these materials. © 2003 American Institute of Physics.

ROSSANA HERNANDEZ

Electrónica del estado solido

http://apl.aip.org/applab/v83/i18/p3731_s1?isAuthorized=no

---

NITROGEN RESONANT STATES IN DISORDERED GaNxAs1−x STRUCTURES

Clearly demonstrate that the conduction band edge in GaNxAs1−x is being perturbed and pushed downward due to its interaction with a higherlying localized resonant state, centered on the nitrogen atoms. Why, then, has this state not been identified in previous calculations? To answer this question, and to investigate the role of disorder, we extend the tight-binding and two-level model to disordered GaNxAs1−x supercells.

We first consider a set of 1000 atom supercells containing up to 15 randomly distributed N atoms. In these supercells we fit the number, but not the distribution, of NN pairs to the number given statistically, so that each cell contains n isolated N sites and p N-N pairs. For each configuration, we used the GULP molecular relaxation package [48] to calculate the equilibrium positions of all the atoms, using a parameterized valence-force-field model, while using Végard's law to vary the unit cell basis vectors as a(x)=x aGaN+(1−x) aGaAs. The calculated relaxed bond lengths are in good agreement with those obtained by other authors [46] who used an ab initio pseudopotential approach.

In a disordered supercell, we can again try to describe the GaNxAs1−x conduction band edge by a Linear Combination of Isolated Nitrogen Resonant States (LCINS) interacting with the unperturbed conduction band edge, ψc0. For the supercells considered here, we have n resonant basis states, associated with isolated N resonances (i=1−n) and 2p resonant basis states associated with the p N-N pairs and j=1–p, which are even and odd, respectively, about the Ga site at the center of the N-N pair). We write the sp3s* Hamiltonian H of the Ga500Nn+2pAs500− n−2P supercell as H=H0+ΔVN+ΔVNN

ROSSANA HERNANDEZ

Electrónica del estado solido

http://books.google.co.ve/books?hl=es&lr=&id=Bq3UeyqU3jsC&oi=fnd&pg=PR13&dq=Physics+and+Applications+of+Dilute+Nitrides&ots=vxlYBVPkMw&sig=mXEomXJ-8lFE6Tv6mm6ezOW7Ubk#v=onepage&q=Physics%20and%20Applications%20of%20Dilute%20Nitrides&f=false

---

TEN-BAND k・p MODEL FOR DILUTE NITRIDE ALLOYS

The k・p and envelope-function methods are widely applied to study III�V semiconductor heterostructures. The strong interaction between the N resonant states and the conduction band edge means that the conventional eight-band k・p method cannot be applied to GaInNAs and related hetero-structures. We must include the interaction between the N resonant states and the conduction band edge to describe the variation of the (zonecenter) conduction-band-edge energy with N. This leads to a modified ten-band k・p Hamiltonian for GaInNAs, with the modified Hamiltonian giving a good description of the conduction-band dispersion over an energy range at least on the order of 200 meV [49], sufficient for most analyses.

We illustrate this by comparing the band structure of a Ga32As32 and a Ga32N1As31 supercell in Figure 3.8, where the dotted lines show the sp3s* band structure plotted with the spin-orbit interaction Eso set to zero. The GaAs eight-band k・p Hamiltonian reduces to a two-band Hamiltonian for the conduction and light-hole valence bands along the [0,0,1] direction when Eso=0, as illustrated by the thick solid lines, which show the dispersion of these two bands calculated using ψc0 and the light-hole zone-center wavefunction, ψlh0, as the k・p basis states. The k・p matrix elements were found by explicitly evaluating <ψi0|H(kz)|ψj0> using the tight binding Hamiltonian [49].

We must add the nitrogen resonant state ψN0 to the k・p Hamiltonian for Ga32NAs31. The conduction and light-hole band dispersion are then

found by diagonalizing a 3×3 k・p model. The most general form of this 3×3 Hamiltonian includes k-dependent diagonal and off-diagonal matrix elements linking the ψN0, ψc0.

The thick solid lines in Figure 3.8b show the band structure of Ga32NAs31 calculated using Equation 3.8, where we evaluate the matrix elements directly using the tightbinding Hamiltonian. This Hamiltonian gives an excellent fit to the conduction-band dispersion within about 200 meV of the band edge. However, it is notable that the N impurity band in Figure 3.8b does not correspond to a specific higher-lying conduction b and in the supercell.

ROSSANA HERNANDEZ

Electrónica del estado solido

http://link.aip.org/link/?apl/81/2523

---

An Atomistic View of the Electronic Structure of Mixed-Anion III–V Nitrides

The conduction-band splitting into two nonparabolic subbands predicted by the BAC model has been unambiguously observed in GaNxAs1−x and Ga1−yInyNxAs1−x using photomodulation spectroscopy [26,42,43]. Figure 2.4 shows photoreflectance (PR) spectra recorded from GaNxAs1−x samples. The PR spectrum of GaAs (x=0) exhibits two sharp derivativelike spectral features corresponding to the transition from the top of the valence band to the bottom of the conduction band (E0 transition), and the transition between the spin-orbit split-off band and the conduction-band minimum (E0+Δ0 transition). For N-containing samples, in addition to the PR spectral.

features related to the transition across the fundamental band gap (E− transition) and the transition from the top of the spin-orbit split-off valence band to the bottom of the conduction band (E−+Δ0 transition), an extra feature (E+) appears at higher energies in the PR spectra. With increasing N concentration, the E− and E−+Δ0 transitions shift to lower energy, and the E+ transition moves in the opposite direction. Shown in Figure 2.5 are the E− and E+ transition energies in Ga1−yInyNxAs1−x as a function of N concentration reported by several different groups [41,43–45]. The nonlinear dependence of the transition energies on N concentration can be well described by the BAC model using a coupling constant V=2.7 eV.

It is also worth noting that, as shown in Figure 2.4, the spin-orbit splitting energy Δ0 is equal to ≈0.34 eV for all the measured samples and does not depend on N content. The results demonstrate that incorporation of N into GaAs and GaInAs affects mostly the conduction band and has a negligible effect on the electronic structure of the valence band. Using Equation 2.9, one can obtain a simple relationship between the subband-

ROSSANA HERNANDEZ

electrónica del estado solido

http://www.routledge.com/books/details/9781591690191/

---

Interband Optical Transitions

The conduction-band splitting into two nonparabolic subbands predicted by the BAC model has been unambiguously observed in GaNxAs1−x and Ga1−yInyNxAs1−x using photomodulation spectroscopy [26,42,43]. Figure 2.4 shows photoreflectance (PR) spectra recorded from GaNxAs1−x samples. The PR spectrum of GaAs (x=0) exhibits two sharp derivativelike spectral features corresponding to the transition from the top of the valence band to the bottom of the conduction band (E0 transition), and the transition between the spin-orbit split-off band and the conduction-band minimum (E0+Δ0 transition). For N-containing samples, in addition to the PR spectral.

features related to the transition across the fundamental band gap (E− transition) and the transition from the top of the spin-orbit split-off valence band to the bottom of the conduction band (E−+Δ0 transition), an extra feature (E+) appears at higher energies in the PR spectra. With increasing N concentration, the E− and E−+Δ0 transitions shift to lower energy, and the E+ transition moves in the opposite direction. Shown in Figure 2.5 are the E− and E+ transition energies in Ga1−yInyNxAs1−x as a function of N concentration reported by several different groups [41,43–45]. The nonlinear dependence of the transition energies on N concentration can be well described by the BAC model using a coupling constant V=2.7 eV.

It is also worth noting that, as shown in Figure 2.4, the spin-orbit splitting energy Δ0 is equal to ≈0.34 eV for all the measured samples and does not depend on N content. The results demonstrate that incorporation of N into GaAs and GaInAs affects mostly the conduction band and has a negligible effect on the electronic structure of the valence band. Using Equation 2.9, one can obtain a simple relationship between the subband-

ROSSANA HERNANDEZ

electrónica del estado solido

http://www.routledge.com/books/details/9781591690191/

---

Effects of the Higher Conduction-Band Minima

Alternative interpretations of some of the observed effects discussed above, such as the appearance of E+ transition and the pressure dependence of the E− and E+ transitions, were also proposed [49–56]. For instance, it has been argued that the observed changes in the conduction band structure are a result of interactions between states originating from the extended states of the Γ, L, and/or X conduction-band minima. It has been argued that incorporation of N breaks the crystal symmetry and splits the degenerate L and X minima into the a1 and t2 states. The a1 states strongly interact with the states at the Γ minimum, leading to a downward shift of the conduction-band edge. The close proximity of the L

minimum energy at EV+1.705 eV to the energy of the localized N-state EN≈EV+1.65 eV was invoked in the argument that interaction with either of these states could be responsible for the E+ and E− transitions [44]. It was also proposed that the impuritylike band of interacting nitrogen pairs and cluster states is responsible for the downward shift of the conduction-band edge in GaNAs alloys [55,56]. Several groups have studied the N-induced effects on the L conduction-band edges by measuring the E1 (Λ4v,5v−Λ6c) transition near the L points of the Brillouin zone in GaNxAs1−x using different experimental methods [33,35,57,58]. Figure 2.11 shows PR spectra associated with the E1 transitions from the L4,5 valence-band edge to the L6

conduction-band minima in several GaNxAs1−x samples. A small increase of the E1 transition energy with increasing x has been observed relative to the E1 transition at 2.925 eV in GaAs.

is, within the experimental uncertainties, the same as downward shift of the E− transition, which represents the fundamental band-gap reduction in the GaNAs samples. This salient feature is a typical characteristic of two-level anticrossing interaction, in which the upward shift of the upper state and the downward shift of the lower state are exactly the same in magnitude. The much slower upward shift of the E1 compared with the E+ transition energy rules out the possibility that the E+ can be associated with N-induced Γv−Lc transition. The slow, monotonic increase of the E1 transition energy with N concentration and the lack of a splitting of the L-band edge are also in disagreement with the theoretical calculations [51] attributing the E+ spectral feature to transitions from Γv to configuration-weighted average of nitrogenlike a1(N) and L-like a1(Lc) states.

To further elucidate the role of the higher energy minima, the effects of N on the band structure of Ga1−yAlyAs alloys were investigated [58]. In these alloys, the Γ band-edge minimum shifts from about 0.5 eV below in GaAs to slightly over 0.5 eV above the X conduction-band minima in AlAs. The large relative energy shift is expected to strongly affect the strength ofthe interaction between those two minima. Ga1−yAlyNxAs1−x alloys used in the study were synthesized by implanting nitrogen ions into MOVPE-grown AlyGa1−yAs epitaxial films on GaAs substrates followed by postimplantation thermal annealing. Energy positions of the experimentally observed E+ and E− transitions for a GaN0.0085As0.9915 and fou Ga1−yAlyNxAs1−x samples are shown in Figure 2.13. The dependencies of the energies of the Γ, L, and X conduction-band minima on the Al content in AlyGa1−yAs are shown inthe figure. The band-gap E0 measured in the as-grown wafers was used to determine the Al concentration of the Ga1−yAlyAs epitaxial films. The inset in Figure 2.13 shows a comparison of the PR spectra between an as-grown Al0.35Ga0.65As sample and an N+- implanted Al0.35Ga0.65NxAs1−x sample.

ROSSANA HERNADEZ

electrnica del estado solido

http://www.emrs-strasbourg.com/index.php?option=com_content&task=view&Itemid=114&id=313

---