AlN Si Research Articles

Silicon diffusion in AlN

In this study, we investigate the diffusion of Si donors in AlN. Amorphous Si1−xNx sputtered on the surface of bulk AlN with low dislocation density is used as a Si source. The diffusion experiments are conducted through isochronal and isothermal annealing in a protective N2 atmosphere at temperatures between 1500 and 1700 °C. The Si depth profiles measured by secondary ion mass spectrometry exhibit a convex box-like shape with a steep diffusion front. These concentration profiles are best described with a diffusion coefficient that depends on the square of local Si concentration. From the characteristic box-shaped Si profiles, we conclude that diffusion of Si in AlN is mediated by singly negatively charged dopant–vacancy pairs SiAlVAl−. The strong concentration dependence of Si diffusion is due to the electric field associated with the incorporation of Si donors (SiAl+1) on substitutional Al lattice sites and reflects that Si is fully electrically active at diffusion temperature. The experimentally obtained extrinsic Si diffusion coefficient is reduced to intrinsic doping conditions. The temperature dependence of Si diffusion for intrinsic conditions is described by an activation enthalpy of (10.34±0.32)eV and a pre-exponential factor of 235−203+1485cm2s−1. The migration enthalpy of the donor–vacancy pair SiAlVAl− is estimated to be around 3.5 eV. This estimation is based on the activation enthalpy of the transport capacity of SiAlVAl− and theoretical results concerning the formation energy of negatively charged vacancies on Al-sites in AlN.

Stable ferromagnetic state in Si-doped AlN with cation vacancies: Ab-initio study

The magnetic property of Si-doped AlN with Al-vacancy is studied using first principles calculations based on spin polarized density functional theory. The Si dopant alone does not introduce the magnetic moment in AlN. However, the doping of Si in AlN reduces the formation energy caused by Al-vacancy, and stabilizes the spin polarized state. The magnetic moments are mainly localized on N atoms surrounding the defect. The strong ferromagnetic state is obtained in AlN due to the combined role of Al-vacancy and Si-dopant.

Synthesis of three-dimensional AlN–Si3N4 branched heterostructures and their photoluminescence properties

Three-dimensional AlN–Si3N4 branched heterostructures are synthesized via extended vapor–liquid–solid and vapor–solid growth of one-dimensional Si3N4 nanostructures and AlN nanocones successively.

Modification of dislocation behavior in GaN overgrown on engineered AlN film-on-bulk Si substrate

The changes that the AlN buffer and Si substrate undergo at each stage of our substrate engineering process, previously shown to lead to a simultaneous and substantial reduction in film crack density and dislocation density in overgrown GaN, are presented. Evidence of ion-implantation assisted grain reorientation for AlN islands coupled with physical isolation from the bulk Si substrate prove to be the dominating driving forces. This is further emphasized with x-ray diffraction analysis that demonstrates a reduction in the in-plane lattice constant of AlN from 3.148 Å to 3.113 Å and a relative change in rotation of AlN islands by 0.135° with regard to the Si substrate after substrate engineering. Misfit dislocations at the AlN-Si interface and disorder that is normally associated with formation of amorphous SiNx at this interface are considered to be two of the major contributors to dislocation nucleation within overgrown GaN. Following our technique, the disappearance of disorder at the AlN-Si interface is observed. Extensive ellipsometry and transmission electron microscopy suggests that larger AlN islands with a smoother surface morphology could further reduce the dislocation density below that previously reported. A 1.2 μm GaN layer deposited on an AlN buffer with larger islands and smoother morphology exhibits a 14× reduction in surface pit density after undergoing the ion-implantation assisted substrate modification technique.

Calculation of an n<SUP align="right">+</SUP>pSi diode collection efficiency and comparison with experiments

A theoretical model accounting for the experimental results of electron beam induced current (EBIC) collection efficiency on Al-n Si diodes and Si n+p cells, subject to various heat treatments, has been put forward. Twelve samples have been investigated, i.e., monocrystals and polycrystals, oxidised and non-oxidised, and this, at three (03) different temperatures (800°C during 60 mn, 850°C during 12 mn and 900°C during 7 mn). The framework adopted has highlighted the importance of some parameters, such as the minority carrier diffusion length, their superficial recombination velocity, together with a carrier trapping velocity. The mutual influences of the material microstructure under study, the temperature and the treatment time have also been analysed.

Shallow donor and DX states of Si in AlN

In unintentionally Si-doped AlN, the electron paramagnetic resonance (EPR) spectrum of the Si shallow donor (g=1.9905) was observed in darkness at room temperature. The temperature dependence of the EPR signal suggests that Si in AlN is a DX center with the DX− state lying at ∼78 meV below the neutral shallow donor state. With such relatively small thermal activation energy, Si is expected to behave as a shallow dopant in AlN at normal device operating temperatures.

Complex phase compositions in nanostructured coatings as evidenced by photoelectron spectroscopy: The case of Al–Si–N hard coatings

The chemical state evolution of the Al–Si–N thin films at various Si contents is investigated by x-ray photoelectron spectroscopy (XPS). The detailed evolution of the Al 2p, Si 2p, and N 1s photoelectrons line positions and widths are used to identify different chemical environments as the Si content is changed. The results are compared to x-ray diffraction (XRD) data that indicate the formation of a two-phase Al1−xSixN/SiNy composite when the solubility limit of 6 at. % of Si in AlN is exceeded. In contrast to XRD data, no particular effect is observed in the XPS data at the solubility limit of Si. Instead, two compositional regions can be identified that are separated by a distinct change in the evolution of core level binding energy differences and chemical shifts at about 10–15 at. % of Si. This silicon concentration is identified as the onset of the formation of a SiNy intergranular phase that is a few monolayers thick, having a chemical bonding similar to that in bulk silicon nitride. The observed changes in the XPS data coincide well with the structural changes in the material at different silicon contents. The unambiguous identification of phases, especially of minority phases from XPS data, is, however, not possible.

Doping efficiency and segregation of Si in AlN grown by molecular beam epitaxy

AbstractThe growth of Si‐doped 2H‐AlN thin films by plasma‐induced molecular‐beam epitaxy is reported. We have found that Si positively affects the epitaxy being an effective surfactant for AlN growth with a remarkable impact on the crystal quality. It was also observed that heavy doping conditions result in the volume segregation of Si at the threading dislocation network and in the formation of an amorphous (AlO)(SiO)N cap layer caused by post‐growth surface oxidation of the accumulated Al and segregated Si. (© 2006 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)

Dielectric properties of AlN film on Si substrate

AlN films were deposited on p-Si(1 1 1) substrates by a low-pressure metal–organic chemical vapor deposition (MOCVD) system and an Al–AlN–Si (MIS) capacitor was then prepared with the AlN layer as the insulator. With the capacitance–voltage (C–V) curves of Al–AlN–Si capacitors measured at 50 Hz–10 MHz, we found the frequency dependence of the capacitance of the accumulation region, that is, the capacitance of the AlN layer. It is observed that the relative dielectric constant (er) of AlN declined from 20.7 to 1.2 with increasing frequency. The peak of dielectric loss tangent (tan δ) was observed at a frequency of 1.5 MHz. With the Debye equation and the Cole–Cole empirical formula, the relaxation time (τ) was fitted to be about 10−7 s. From these results and works of others, it was concluded that this dielectric behavior was caused by dipolar relaxation relating to N vacancies in the AlN layer.

Orientation of aluminum nitride films grown with hyperthermal molecular beams

Unskimmed, pulsed hyperthermal molecular beams of trimethylaluminum (TMA) and ammonia were used to grow films of aluminum nitride. The crystal structure of the films was determined by X-ray diffraction (XRD). AlN was deposited on Si(1 0 0) and AlN–Si. The AlN–Si samples consist of 1 μm layer of MOCVD-grown polycrystalline AlN on Si(1 0 0) substrates. The resulting film orientation on Si(1 0 0) is with the AlN( 1 0 1 ̄ 0 ) plane parallel to the substrate plane. On AlN–Si substrates, the AlN( 1 0 1 ̄ 0 ) plane is still prevalent. The effect of varying the precursors’ azimuthal angle of incidence, α, was also investigated. AlN was grown on Si(1 0 0) at α=0, 30, 60°. For α>0, the AlN( 1 0 1 ̄ 0 ) and AlN( 1 0 1 ̄ 1 ) planes are present. At α=60°, weak X-ray diffraction peaks indicate the films have short-range order. For the α examined, the average growth rate is approximately 0.35 μm/h. AES was used for chemical analysis of films.

Interpreting the nonideal reverse bias C-V characteristics and importance of the dependence of Schottky barrier height on applied voltage

This work presents an attempt related to the charging behaviour of interface states to the nonideal forward bias current-voltage ( I-V) and the reverse bias capacitance-voltage ( C-V) characteristics of AlnSi Schottky barrier diodes. The diode showed nonideal I-V behaviour with an ideality factor of 1.50 and was thought to have a metal-interface layer-semiconductor configuration. Considering that the interface states localized at the interfacial layer-semiconductor interface are in equilibrium with the semiconductor, the energy distribution of the interface states was exactly determined from the forward bias I-V characteristics by taking into account the bias dependence of the effective barrier height, θ e. The determination of the intercept voltage and interface state density was made by means of a simple interface charge model which has been developed in detail. The I-V characteristics were used for determining the voltage dependence of the barrier height. Although the change in barrier height with applied biasis small, it is important for exactly determining the shape of the interface state density distribution curve. At a frequency of 500 kHz, the nonlinear reverse bias C −2− V plot with the curvature concave downward has been only thought of to be due to the contribution of the capacitance of the interface state charges. It is concluded that the nonlinear nature of C −2− V plots in the frequency range 50–200 kHz has been caused by the interface state charges as well as inversion layer and inversion layer charges. It has been understood by means of the interface state charge model that the C −2− V plots cannot only be interpreted in terms of the contribution of the interface state charges to the device capacitance.

A simple interfacial-layer model for the nonideal I- V and C- V characteristics of the Schottky-barrier diode

An interfacial-layer theory considering the nonequilibrium generation-recombination mechanism of the real interface states for Schottky-barrier diodes is presented. It is shown in the theory that the soft behavior of the reverse I- V characteristics associated with the observed nearly ideal C −2- V characteristics can be modeled in terms of the interfacial-layer capacitance in a consistent manner, and that the nonideality of the forward I- V as well as C −2- V characteristics should be attributed to the transition from the metal—controlled to semiconductor (majority carrier)—controlled occupancy of the interface states. The experimental verification for the theory is done by comparing the density distribution of the interface states in the semiconductor bandgap extracted from the nonideal forward I– V characteristics of the fabricated AlpSi and AlnSi Schottky barrier diodes with those directly measured by the multifrequency admittance methods. Excellent agreement from these comparisons strongly supports the physical validity of our theory.

Phase equilibria in the Si, Al, Be/C, N system

Abstract Phase equilibria studies in parts of the system Si, Al, Be/C, N were carried out at 1760 and 1860°C. In the carbide system Al 4 C 3 SiCBe 2 C seven new compounds were found and characterized by their chemical composition and crystal symmetry. In the nitride system AlNSi 2 N 4 Be 2 N 2 a solid solution between AlN and BeSiN 2 was found to exist besides the already known compounds. The carbonitride system showed solid solution between α-Al 4 SiC 4 and α-Al 5 C 3 N. Furthermore the compound AlN · SiC · Al 4 C 3 was found to exist.

Phase equilibria in the Si,Al,Be/C,N system

Phase equilibria studies in parts of the system Si, Al, Be/C, N were carried out at 1760 and 1860°C. In the carbide system Al 4C 3SiCBe 2C seven new compounds were found and characterized by their chemical composition and crystal symmetry. In the nitride system AlNSi 2N 4Be 2N 2 a solid solution between AlN and BeSiN 2 was found to exist besides the already known compounds. The carbonitride system showed solid solution between α-Al 4SiC 4 and α-Al 5C 3N. Furthermore the compound AlN · SiC · Al 4C 3 was found to exist.

Aluminum-silicon Schottky barriers as semiconductor targets for EBS devices

The ideal semiconductor target for EBS devices is most closely approximated by the Schottky barrier junction, with Al having the highest figure of merit. Accordingly, large area (1·27 mm 2), high-field Al nSi Schottky barrier junctions surrounded by a p- n junction guard-ring have been fabricated in unpassivated, SiO 2 passivated, and Al 2O 3 passivated configurations. The fabrication procedure is described and data pertaining to forward and reverse characteristics, yield and storage characteristics are presented. The Schottky barrier height is characterized by four independent measurements with excellent agreement between the various measurements. The barrier height is in the 0·710–0·760 V range and the typical ideality factor is ≤1·05. The devices exhibit ideal Schottky barrier behavior at least over the −50−100°C temperature range. The current gain of Schottky barriers with different Al layer thicknesses is measured under static electron bombardment over the 1–30 keV energy range. From these measurements, the average energy required to create a hole-electron pair is determined to be 3·44±0·2 eV. Using the gain-energy characteristic, it is possible to identify an optimum accelerating potential for each target. This optimum accelerating potential is in good agreement with that obtained theoretically from consideration of target losses. No junction instability is observed due to electron irradiation for beam current densities up to 0·314 A/cm 2 and target current densities exceeding 1000 A/cm 2. Further, no significant device degradation is observed for target dissipated power densities exceeding 50 kW/cm 2. Under pulse mode dynamic testing, output pulses of 13·1 A with a measured risetime of 0·72 nsec into the 10 Ω optimum load are obtained. When the output risetime is corrected for beam resetime (∼0·60 nsec), an output/risetime of 13·1 A/0·40 nsec is obtained. This is within a factor of 2 of the theoretical output capability of an optimized target; the difference is due to the capacitance of the guard-ring. The amplifier pulse mode efficiency is 85 per cent. The low pass bandwidth is 875 MHz, and the Pf 2 figure is 82 W-GHz 2. In addition, small-signal transconductance and power gains of 4·35 ℧ and 40 dB, respectively, are achieved. No output pulse distortion or pulse height deterioration is noted for pulse widths up to 130 nsec having 13·0 A amplitude under prolonged bombardment of our devices. This indicates that if any insulator charging is occurring, it does not result in any deleterious effects on the target performance.

Schottky barrier height dependence on Si crystal orientation

Planar Al–nSi and PtSi–nSi Schottky barriers with diffused p-type guard rings have been fabricated on (111) and (100) silicon surfaces. The barriers have been produced by sputtering the metals on sputter-etched silicon surfaces. For Al–nSi barriers the barrier height is 0.72 eV on (111) surfaces and 0.81 eV on (100) surfaces, as determined from the forward I-V characteristic. This difference in barrier potential has been confirmed by C-V measurements. No difference in barrier height is found for PtSi–nSi barriers made on (111) and (100) surfaces. A qualitative explanation is given for the experimental results in terms of strong or weak coupling of surface states to the metal.

Preparation and electrical properties of AlAlNSi structures

en