Zincblende Phase GaN Research Articles

Ammonothermal epitaxy of wurtzite GaN using an NH 4I mineralizer

Purely wurtzite phase needle crystals and epitaxial layers of GaN were grown by the ammonothermal method using an NH 4I mineralizer. The inclusion of zincblende phase GaN was effectively eliminated by increasing the growth temperature higher than 500 °C. Accordingly, an approximately 20-μm-thick GaN epitaxial layer was achieved on the Ga-polar face of a c-plane GaN seed wafer at 520 °C. Although the characteristic deep state emission band dominated the room temperature photoluminescence spectrum, the near-band-edge emission of GaN was observed for both the needle crystals and the epitaxial layers. These results encourage one to grow better quality GaN crystals at a high growth rate under high-temperature growth conditions.

Growth and Characteristics of Zinc-Blende and Wurtzite GaN Junctioned Branch Nanostructures

Both wurtzite and zinc-blende phase junctioned GaN nanostructures were synthesized using thermal chemical vapor deposition methods via the vapor−liquid−solid process for the first time. We observed catalyst movement and the regrowth of nanowire zinc-blende phases. This phenomenon was believed to occur in order to reduce the lattice mismatches between the wurtzite phase GaN and Au planes. The growth route could synthesize the c- and h-GaN junctioned nanostructures. The emission values for the zinc-blende phase GaN nanosturtures in cathodoluminescence were shifted a few meV higher than the reported values because the zinc-blende phase GaN epitaxially grew on wurtzite phase GaN without the residual strain.

Molecular beam epitaxy as a method for the growth of freestanding zinc-blende (cubic) GaN layers and substrates

The authors have studied the growth of bulk, freestanding zinc-blende (cubic) GaN layers by plasma-assisted molecular beam epitaxy (PA-MBE). They have established that the best structural properties of freestanding zinc-blende GaN can be achieved with initiation under Ga-rich conditions but without Ga droplet formation. It is difficult to initiate the growth of zinc-blende GaN, but it is even more difficult to sustain the growth of the pure zinc-blende polytype in thick layers without any wurtzite inclusions. In order to grow high quality freestanding cubic GaN layers, it is necessary to maintain the same growth conditions for about 1week. The best quality zinc-blende phase GaN was achieved in the first 10μm of the GaN layers. The authors have produced zinc-blende GaN substrates from our thick bulk GaN layers and they used the side previously attached to the GaAs substrate as the episide of these zinc-blende GaN substrates. They have demonstrated the scalability of the process by growing zinc-blende GaN layers on 2 and 3in. diameter wafers. The growth of freestanding bulk GaN layers has allowed them to refine the value for the lattice parameter of zinc-blende GaN as 4.510±0.005Å. They have demonstrated that the PA-MBE process developed has allowed them to grow freestanding AlxGa1−xN wafers.

Theoretical Study of RF Breakdown in GaN Wurtzite and Zincblende Phase MESFETs

In this paper, we present a comparison of the RF breakdown behaviors of representative wurtzite and zincblende phase GaN MESFET structures based on a theoretical analysis. The calculations are made using a full band ensemble, Monte Carlo simulation that includes a numerical formulation of the impact ionization transition rate. Calculations of the RF breakdown voltages are presented for submicron MESFET devices made from either wurtzite or zincblende phase GaN. The devices are otherwise identical. It is found that the RF-breakdown voltage of the devices increases with increasing frequency of the applied large signal RF excitation.

Comparison between wurtzite phase and zincblende phase GaN MESFETs using a full band Monte Carlo simulation

Cutoff frequency, breakdown voltage, and the transconductance of wurtzite and zincblende phase GaN MESFETs have been calculated using a self-consistent, full band Monte Carlo simulation. The effect of interface states on the device performance is modeled by including uniformly depleted regions at the device surface under the passivation layers. It is found that the drain current increases gradually with increasing drain-source voltage at the onset of breakdown for both phases. The calculated breakdown voltage for the wurtzite device is considerably higher than the breakdown voltage calculated for the zincblende device. On the other hand, the zincblende device is calculated to have higher transconductance and cutoff frequency than the wurtzite device. The higher breakdown voltage of the wurtzite phase device is attributed to the higher density of electronic states for this phase compared to the zincblende phase. The higher cutoff frequency and transconductance of the zincblende phase device is apparently due to the greater electron velocity overshoot for this phase compared to that for the wurtzite phase. The maximum cutoff frequency and transconductance of a 0.1 /spl mu/m gate-length zincblende GaN MESFET are calculated to be 220 GHz and 210 mS/mm, respectively. The corresponding quantities for the wurtzite GaN device are calculated to be 160 GHz and 158 mS/mm.

Full Band Monte Carlo Comparison of Wurtzite and Zincblende Phase GaN MESFETs

The output characteristics, cutoff frequency, breakdown voltage and the transconductance of wurtzite and zincblende phase GaN MESFETs have been calculated using a self-consistent, full band Monte Carlo simulation. It is found that the calculated breakdown voltage for the wurtzite device is considerably higher than that calculated for a comparable GaN zincblende phase device. The zincblende device is calculated to have a higher transconductance and cutoff frequency than the wurtzite device. The higher breakdown voltage of the wurtzite phase device is attributed to the higher density of electronic states for this phase compared to the zincblende phase. The higher cutoff frequency and transconductance of the zincblende phase GaN device is attributed to more appreciable electron velocity overshoot for this phase compared to that for the wurtzite phase. The maximum cutoff frequency and transconductance of a 0.1 μm gate-length zincblende GaN MESFET are calculated to be 220GHz and 210 mS/mm, respectively. The corresponding quantities for the wurtzite GaN device are calculated to be 160GHz and 158 mS/mm.

Analysis of millimeter-wave GaN IMPATT oscillator at elevated temperature

GaN is a suitable material for millimeter-wave high-power IMPATT oscillators because of its superior electronic properties—high breakdown electric fields and high electron saturation velocity. In this paper, millimeter-wave wurtzite phase and zincblende phase GaN IMPATT oscillators at elevated temperature are analyzed by a Read-type large-signal model. The power density of GaN IMPATT devices at millimeter-wave frequencies is two orders magnitude higher than that of conventional GaAs and Si IMPATT devices. The simulations showed that GaN wurtzite phase p+-n single-drift flat-profile IMPATT oscillators at 300 GHz have an efficiency of 11% and an RF power density of 1.6 MW/cm2 when operated at 800 K. ©1999 John Wiley & Sons, Inc. Microwave Opt Technol Lett 23: 257–259, 1999.

Full band Monte Carlo simulation of zincblende GaN MESFET's including realistic impact ionization rates

In this paper, we present the first theoretical study of the breakdown properties of zincblende phase GaN MESFET devices. The calculations are made using a full band, ensemble Monte Carlo simulation that includes a numerical formulation of the impact ionization transition rates. The breakdown voltage, transconductance and cutoff frequency are calculated for the GaN MESFET under two different conditions, with and without semiconductor-oxide interface states. Uniform surface depletion regions model the effect of the interface states. It is found that the breakdown voltage of the zincblende GaN MESFET is less dependent upon the surface depletion conditions than a corresponding GaAs MESFET. It is also found that the drain current increases more gradually with increasing drain-source voltage at the onset of breakdown and that the breakdown voltage of the zincblende GaN MESFET is predicted to be several times larger than that of a comparable GaAs MESFET. The maximum current gain cutoff frequency of a 0.1 /spl mu/m gate length GaN MESFET is calculated to be 230 and 220 GHz, for the non-surface-depleted and the surface depleted devices respectively.

Full Band Monte Carlo Comparison of Wurtzite and Zincblende Phase GaN MESFETs

AbstractThe output characteristics, cutoff frequency, breakdown voltage and the transconductance of wurtzite and zincblende phase GaN MESFETs have been calculated using a self-consistent, full band Monte Carlo simulation. It is found that the calculated breakdown voltage for the wurtzite device is considerably higher than that calculated for a comparable GaN zincblende phase device. The zincblende device is calculated to have a higher transconductance and cutoff frequency than the wurtzite device. The higher breakdown voltage of the wurtzite phase device is attributed to the higher density of electronic states for this phase compared to the zincblende phase. The higher cutoff frequency and transconductance of the zincblende phase GaN device is attributed to more appreciable electron velocity overshoot for this phase compared to that for the wurtzite phase. The maximum cutoff frequency and transconductance of a 0.1 μm gate-length zincblende GaN MESFET are calculated to be 220GHz and 210 mS/mm, respectively. The corresponding quantities for the wurtzite GaN device are calculated to be 160GHz and 158 mS/mm.

Comparison of Electron and Hole Initiated Impact Ionization in Zincblende and Wurtzite Phase Gallium Nitride

ABSTRACTIn this paper, we present the first calculations of the electron and hole impact ionizatioi coefficients for both wurtzite and zincblende phase GaN as a function of the applied electrii field. The calculations are made using an ensemble Monte Carlo simulator including the ful details of the conduction and valence bands derived from an empirical pseudopotentia calculation. The interband impact ionization transition rates for both carrier species an determined by direct numerical integration including a wavevector dependent dielectric function It is found that the electron and hole ionization coefficients are comparable in zincblende Ga> at an applied field of ∼ 3 MV/cm, yet vary to a slight degree at both higher and lower applied field strengths. In the wurtzite phase, the electron and hole coefficients are comparable at hig] fields but diverge at lower applied fields. The most striking result is that the ionization rates an predicted to be substantially different for both carrier species between the two phases. It i predicted that the ionization rates for both carrier species in the zincblende phase are significanti; higher than in the wurtzite phase over the full range of applied fields examined.

Theoretical Prediction of Zinc Blende Phase GaN Avalanche Photodiode Performance Based on Numerically Calculated Electron and Hole Impact Ionization Rate Ratio

AbstractIn this paper, we present the first calculations of the electron and hole initiated interband impact ionization rate in zinc blende phase GaN as a function of the applied electric field strength. The calculations are performed using an ensemble Monte Carlo simulator including the full details of the conduction and valence bands along with a numerically determined, wave-vector dependent interband ionization transition rate determined from an empirical pseudopotential calculation. The first four conduction bands and first three valence bands, which fully comprise the energy range of interest for device simulation, are included in the analysis. It is found that the electron and hole ionization rates are comparable over the full range of applied electric field strengths examined. Based on these calculations an avalanche photodiode, APD, made from bulk zinc blende GaN then would exhibit poor noise and bandwidth performance. It should be noted however, that the accuracy of the band structure employed and the scattering rates is presently unknown since little experimental information is available for comparison. Therefore, due to these uncertainties, it is difficult to unequivocally conclude that the ionization rates are comparable.