Mg-doped GaN Layers Research Articles

Monolithic E/D-mode P-GaN/AlGaN/GaN HFETs using reactivation annealing process

A reactivation annealing process was developed to restore the deactivated p-type characteristics of an Mg-doped p-type GaN layer subsequent to exposure to the SiNx passivation step in a p-GaN gated AlGaN/GaN heterojunction field-effect transistor (HFET). The deactivation, induced by the formation of H–Mg complexes during the passivation process, necessitates an additional activation annealing procedure. Notably, it was observed that the passivation film on the p-GaN layer must be removed through etching prior to annealing, aiming to eliminate H atoms from H–Mg complexes. The reactivation annealing process was conducted at 800 °C. Employing this reactivation approach, we successfully demonstrated monolithic enhancement- and depletion-mode p-GaN/AlGaN/GaN HFETs.

Suppression of threshold voltage shift due to positive bias stress in GaN planar MOSFETs by post-deposition annealing

Threshold voltage instability (shift) due to positive bias stress in GaN planar-gate MOSFETs was investigated. Gate dielectric (SiO2) was formed by remote-plasma-assisted CVD on homoepitaxial Mg-doped p-type GaN layers with Si-implanted n-type source and drain regions. The threshold voltage shift of 5.8 V was observed after a stress voltage of 30 V for a sample without post-deposition annealing (PDA). The threshold voltage shift was significantly reduced to 1.4 V for a sample with PDA (800 °C for 30 min). Stress time dependences up to 6000 s were measured, revealing that the main origin of the threshold voltage shift is electron trapping into near interface traps (NITs). These results suggest that PDA is effective for the reduction of the NITs.

(Invited) Effect of Oxygen on Activation Annealing of Mg-Doped GaN

The effects of N2 and O2:N2 (1:1) as ambient gases during activation annealing of Mg as p-type doping of GaN grown by MOCVD are investigated. The purpose is to understand the mechanisms involved and especially the impact of O2 on the resulting free hole concentration and hole mobility. The addition of O2 to the ambient gas during annealing is very effective in reducing the H level of the Mg-doped GaN layer, especially at lower activation temperatures, but the maximum achievable hole concentration and mobility, as determined by Hall characterization, is still higher with pure N2.The difference between activation in pure N2 or O2:N2 (1:1) ambient a is explained by an in-diffusion of O to the GaN layer acting as n-dopant and thus giving rise to a compensation effect. It is found that to a large degree only the Mg-H complexes at substitutional (MgGa), i.e., the electrically active acceptor sites that provide free holes, are activated by annealing with N2 only as ambient gas, while annealing with O2:N2 (1:1) also dissociates electrically inactive Mg-H complexes resulting in much less residual H. Thus, the residual H level in relation to the Mg level after activation annealing with N2 only may provide a representative measure of the resulting free hole concentration of the Mg-doped GaN layer.An additional factor to be taken into account for in order to maximize the hole concentration is the growth pressure. At the lower growth pressure of 150 mbar compared to higher growth pressure of 300 mbar, an increasing carbon incorporation leads to a compensation effect drastically reducing the free hole concentration while the mobility is unaffected. Conclusions Addition of O2 during RTP is very effective in reducing the H level however, the maximum hole concentration and mobility are still higher with pure N2.Mg-H complexes at substitutional (MgGa) electrically active acceptor are preferentially dissociated by annealing with N2 as ambient gas.By employing a higher growth pressure, carbon incorporation that leads to a compensation effect drastically reducing the free hole concentration can be minimized.By activation annealing with N2 the residual H level provides a measure of the free hole concentration. Figure 1

Acceptor activation of Mg-doped GaN—Effects of N2/O2 vs N2 as ambient gas during annealing

Here, we investigate the effects of O2:N2 (1:1) as ambient gas as compared with pure N2 during activation annealing of Mg as p-type doping in GaN layers grown by MOCVD. The purpose is to understand the impact of O2 on the resulting free hole concentration and hole mobility using SIMS, XRD, STEM, AFM, and Hall effect measurements. Even though the presence of O2 in the ambient gas during annealing is very effective in reducing the H level of the Mg-doped GaN layers, the maximum achievable hole concentration and mobility is still higher with pure N2. The differences are explained by an in-diffusion of O to the GaN layer acting as n-dopant and, thus, giving rise to a compensation effect. The Mg–H complexes at substitutional (MgGa), i.e., the electrically active acceptor sites that provide free holes, are preferentially activated by annealing with N2 only as ambient gas, while annealing with O2:N2 (1:1) also dissociates electrically inactive Mg–H complexes resulting in much less residual H. At the lower growth pressure of 150 mbar compared to higher growth pressure of 300 mbar, an increasing carbon incorporation leads to a compensation effect drastically reducing the free hole concentration while the mobility is unaffected.

Understanding the influence of physical properties on the mechanical characteristics of Mg-doped GaN thin films

This study investigates the impact of Mg doping concentration in GaN thin films on their physical and mechanical properties, specifically on their conductivity transition, luminescence, hardness, Young's modulus, and creep behavior. Mg-doped GaN layers were grown on sapphire substrates using the MOCVD technique, with trimethylgallium (TMG) and bis(cyclopentadienyl) magnesium (Cp2Mg) as precursors for Ga and Mg, respectively. The TMG flow rate was set at 20 μmol/min, while the Mg flow rate was adjusted by tuning the temperature of the thermostatic bath of Cp2Mg. Three different Cp2Mg flow rates: 2, 7 and 9 μmol/min were utilized to explore their effects on the physical and mechanical properties of the grown epilayers. The results demonstrate that an increase in the Mg concentration leads to a conductivity transition from n-type to p-type, with the highest hole concentration (4.5 ± 0.2) × 1017 cm−3 and blue luminescence attained with Mg concentration equal to 2.1 × 1019 atoms/cm3. Moreover, the hardness, Young's modulus and the intensity of compressive stress increase with the Cp2Mg flow rate due to the evolution of pre-existing dislocation density and point defects incorporation. Additionally, this study evaluates the creep behavior of the Mg-doped GaN thin films. It shows that both the creep stress exponent and the maximal creep depth decrease with the increase of the Cp2Mg flow rate. This is attributed to dislocation glides and climbs governing the creep mechanism. Accordingly, this investigation highlights the importance of understanding the relationship between physical properties and mechanical characteristics of Mg-doped GaN thin films, which have potential implications for various technological applications.

Mechanical Properties and Creep Behavior of Undoped and Mg-Doped GaN Thin Films Grown by Metal–Organic Chemical Vapor Deposition

This paper investigates the mechanical properties and creep behavior of undoped and Mg-doped GaN thin films grown on sapphire substrates using metal–organic chemical vapor deposition (MOCVD) with trimethylgallium (TMG) and bis(cyclopentadienyl)magnesium (Cp2Mg) as the precursors for Ga and Mg, respectively. The Mg-doped GaN layer, with a [Mg]/[TMG] ratio of 0.33, was systematically analyzed to compare its mechanical properties and creep behavior to those of the undoped GaN thin film, marking the first investigation into the creep behavior of both GaN and Mg-doped GaN thin films. The results show that the incorporated [Mg]/[TMG] ratio was sufficient for the transition from n-type to p-type conductivity with higher hole concentration around 4.6×1017 cm−3. Additionally, it was observed that Mg doping impacted the hardness and Young’s modulus, leading to an approximately 20% increase in these mechanical properties. The creep exponent is also affected due to the introduction of Mg atoms. This, in turn, contributes to an increase in pre-existing dislocation density from 2 × 108 cm−2 for undoped GaN to 5 × 109 cm−2 for the Mg-doped GaN layer. The assessment of the creep behavior of GaN and Mg-doped GaN thin films reveals an inherent creep mechanism governed by dislocation glides and climbs, highlighting the significance of Mg doping concentration in GaN thin films and its potential impact on various technological applications.

Selectively patterned Mg-doped GaN by SiNx-driven hydrogen injection

We demonstrate a method to achieve selectively patterned Mg-doped GaN layers using hydrogen drive-in through plasma-enhanced chemical vapor deposition (PECVD) silicon nitride (SiNx) films. Activated Mg-doped GaN layers were selectively deactivated by patterned PECVD SiNx films with low-temperature annealing and showed high-resistive behavior. Spatially resolved photoluminescence measurements were used to optically verify the deactivation of Mg acceptors and showed distinct features corresponding to activated and deactivated Mg in GaN. The method suggested here provides a simple and effective method to achieve patterned Mg-doped GaN regions without thermal and plasma damage, which could cause degradation of device performance. The proposed method could provide a way to achieve future high-performance GaN lateral and vertical devices that rely on laterally patterned doping.

Laser-induced activation of Mg-doped GaN: quantitative characterization and analysis

We investigate the effectiveness of laser-induced treatment as compared to rapid-thermal annealing (RTA) for the activation of p-type dopant in Mg-doped GaN layers. The study is based on a wide set of analytical techniques, including resistivity measurements, atomic force microscopy (AFM), scanning emission microscopy, dynamic secondary ion mass spectroscopy (SIMS), time-of-flight (TOF) SIMS and energy dispersive x-ray (EDX) spectroscopy in combination with scanning transmission electron microscopy (STEM). Samples are treated at different energy densities and in different atmospheres, to provide a comprehensive overview of the topic. The analysis is carried out on GaN-on-Si samples, to demonstrate the effectiveness of the treatment even in presence of high threading dislocation densities. The original results presented in this paper indicate that: (a) laser treatment is an effective process for activating the p-type dopant in Mg-doped GaN layers; even at low irradiation energy densities (400 mJ cm−2) the laser treatment can effectively activate the Mg doping, with the best resistivity results obtained (around 1.5 Ωcm) comparable with those obtained by optimized RTA; (b) resistivity varies with temperature with activation energy E a = 0.14 eV, which is compatible with the MgGa acceptor in GaN; (c) TOF-SIMS, AFM, EDX-STEM analysis indicates that the laser treatment does not modify the concentration profile of magnesium and surface roughness for low and moderate laser energy densities; changes are detected only for energy densities above 600 mJ cm−2, for which a significant degradation of the surface is revealed. The experimental evidence collected within this paper provide an accurate assessment of the process conditions for effective laser activation of Mg-doped GaN, thus allowing the fine-tuning required for selective activation and for industrial applications.

Inhibition of Mg activation in p-type GaN caused by thin AlGaN capping layer and impact of designing hydrogen desorption pathway

Thermal activation of a Mg-doped GaN layer with a thin AlGaN capping layer was investigated by means of systematic electrical characterization. Two-dimensional electron gas generated at the AlGaN/GaN interface greatly inhibited activation of the underlying Mg-doped GaN layer. This finding is attributed to the charge states of the interstitial hydrogen atoms being released from the Mg–H complexes and their diffusivity being dependent on the Fermi level position in GaN-based heterostructures. Reasonable electrically active acceptor concentration in Mg-doped GaN was achieved by designing a hydrogen desorption pathway.

Electrical properties and structural defects of p-type GaN layers grown by halide vapor phase epitaxy

The electrical properties and structural defects of p-type GaN layers with Mg concentrations from 8.0 × 1018 to 8.3 × 1019 cm−3 grown by halide vapor phase epitaxy (HVPE) were investigated. In all samples, p-type conduction was confirmed at room temperature. The hole concentration at room temperature decreased in a heavily Mg-doped sample. By analyzing the results of Hall-effect measurements at various temperatures, the acceptor concentration decreased in a heavily Mg-doped sample, whereas the compensating donor concentration increased. These results affect the decrease in the hole concentration. The hole mobility decreased with increasing acceptor concentration. In the heavily Mg-doped sample, pyramidal inversion domains (PIDs) were formed. The size of each PID in an HVPE-grown sample is in good agreement with that Mg-doped GaN layers grown by metalorganic vapor phase epitaxy (MOVPE). Thus, the formation mechanism of PIDs in HVPE-grown samples is possibly the same as that in MOVPE-grown samples. Energy-dispersive X-ray spectroscopy shows that Mg atoms accumulate in PIDs, which suggests that Mg atoms in PIDs are electrically inactive, inhibiting the increase in the acceptor concentration. These results are useful guidelines for fabricating p-type GaN layers with higher hole concentrations by HVPE.

Analysis of channel mobility in GaN-based metal-oxide-semiconductor field-effect transistors

The factors limiting channel mobility in AlSiO/p-type GaN-based metal-oxide-semiconductor field-effect transistors (MOSFETs) were systematically investigated. MOSFETs with various thin interfacial layers (ILs) between Al0.78Si0.22Oy films and Mg-doped GaN layers were prepared and found to exhibit different channel mobilities. The maximum effective mobility showed a significant correlation with the threshold voltage (Vt) and the hysteresis (ΔVt) in the transfer characteristics of these devices, such that the mobility decreased with increasing Vt and ΔVt. This effect can be explained by electron capture in and emission from border traps situated near the conduction band minimum for GaN. The insertion of a 3-nm-thick SiO2 IL drastically enhanced the effective mobility and simultaneously reduced the ΔVt value. Hall effect measurements with an applied gate voltage were used to determine the mobility of free electrons while excluding electrons captured in the border traps. The Hall effect mobility was much higher than the effective mobility, indicating that mobility was in fact reduced by the capture of electrons by the border traps. The ratio of electrons captured by border traps to the overall electrons induced by a gate bias was greatly lowered in a MOSFET incorporating a SiO2 IL. When a high vertical electric field of approximately 1 MV/cm was present in the device channel, the Hall effect mobility was slightly increased following the insertion of an IL. These results suggest that the IL reduced the interfacial roughness and/or affected the screening out of scattering due to potential fluctuations of the AlSiO.

Suppression of the regrowth interface leakage current in AlGaN/GaN HEMTs by unactivated Mg doped GaN layer

The leakage current caused by the Si pileup at the regrowth interface of AlGaN/GaN high electron mobility transistors (HEMTs) is significantly suppressed by the semi-insulating Mg-doped GaN layer. Mg is unintentionally doped and can be originated from the graphite susceptor of metal organic vapor phase epitaxy. Before regrowth of the AlGaN/GaN heterostructure, the GaN template is treated with hydrochloric acid (HCl) and hydrogen peroxide/potassium hydroxide (H2O2/KOH) solution to promote the adsorption of Mg on the GaN surface. The Mg-doped GaN channel layer is highly resistive due to the passivation of hydrogen. The p–n junction formed by the weakly p-type Mg-doped GaN layer and the n-type Si pileup layer depletes the excess electrons at the regrowth interface. As a result, the off-state drain leakage current of the HEMT device can be decreased by two orders of magnitude at 40 V.

Mapping the growth of p-type GaN layer under Ga-rich and N-rich conditions at low to high temperatures by plasma-assisted molecular beam epitaxy

We present a detailed study of Mg acceptor, p-type doping of GaN grown by plasma-assisted molecular beam epitaxy in both Ga-rich and N-rich conditions at low (∼580 °C) to high growth temperatures (∼740 °C). A growth map is constructed using results from a broad range of growth conditions, which shows the dependence of Mg incorporation and surface roughness on the III/V ratio and growth temperatures. Detailed secondary ion mass spectroscopy and atomic force microscopy studies confirmed that N-rich conditions are favorable for significantly higher Mg-incorporation efficiency (∼80%), whereas the Ga-rich growth condition is preferable for achieving a smooth surface morphology (root mean square roughness: ∼1–2 nm) with poor Mg incorporation. The room temperature Hall measurement confirms that the hole concentration in the range of ∼7 × 1017 to 2 × 1019 cm−3 can be achieved in Ga-rich and N-rich conditions, respectively, at a fixed Mg flux depending on the growth conditions. Our detailed study provides a proper guideline for realizing an efficient Mg-doped GaN layer and is applicable, in principle, to different nitride-based electronic and photonic devices.

Quantitative investigation of the lateral diffusion of hydrogen in p-type GaN layers having NPN structures

This work quantitatively examined lateral diffusion of hydrogen in the (0001) in-plane direction in a Mg-doped p-type GaN layer sandwiched between n-type layers, based on the preparation and annealing of circular mesa structures having different radius values. Capacitance–voltage analyses and secondary ion mass spectrometry confirmed a decrease in the effective acceptor concentration along with an increase in the residual hydrogen concentration with increasing radius values, indicating hydrogen desorption through the sidewall and lateral diffusion. Considering the surface barrier to desorption, we estimated a lateral diffusion coefficient of 7 × 10−8 cm2 s−1 at 1123 K, reasonably reproducing the effect of the mesa radius.

Ultraviolet-A LED Based on Quantum-Disks-In-AlGaN-Nanowires—Optimization and Device Reliability

Group-III nitride-based ultraviolet (UV) quantum-disks (Qdisks) nanowires (NWs) light-emitting diodes grown on silicon substrates offer a scalable, environment-friendly, compact, and low-cost solution for numerous applications in solid-state lighting, spectroscopy, and biomedical. However, the internal quantum efficiency, injection efficiency, and extraction efficiency need to be further improved. The focus of this paper encompasses investigations based on structural optimization, device simulation, and device reliability. To optimize a UV-A (320–400 nm) device structure, we utilize the self-assembled Qdisks NWs with varying Qdisks thickness to study carrier separation in active-region and implement an improved p-contact-layer to increase the output power. By simulation, we found a 100× improvement in the direct recombination rate for samples with thicker Qdisks thickness of ∼5 monolayers (MLs) compared to the sample with ∼2 MLs-thick Qdisks. Moreover, the sample with graded top Mg-doped AlGaN layer in conjunction with thin Mg-doped GaN layer shows 10× improvement in the output power compared to the samples with thicker top Mg-doped GaN absorbing contact layer. A fitting with ABC model revealed the increase in nonradiative recombination centers in the active region after a soft stress test. This paper aims to shed light on the research efforts required for furthering the UV NWs LED research for practical applications.

Influence of residual carbon impurities in a heavily Mg-doped GaN contact layer on an Ohmic contact.

The influence of residual carbon impurities incorporated into a heavily Mg-doped GaN layer has been studied systematically according to the relation between the carbon concentration and specific contact resistance. Furthermore, the results of temperature-dependent current-voltage characteristics and the photoluminescence spectra indicate that a proper concentration of residual carbon impurities can improve the performance of Ohmic contact by introducing deep-level defects to enhance the variable-range-hopping conduction.

All metalorganic chemical vapor phase epitaxy of p/n-GaN tunnel junction for blue light emitting diode applications

We report on III-Nitride blue light emitting diodes (LEDs) comprising a GaN-based tunnel junction (TJ) all realized by metalorganic vapor phase epitaxy in a single growth process. The TJ grown atop the LED structures consists of a Mg-doped GaN layer and subsequently grown highly Ge-doped GaN. Long thermal annealing of 60 min at 800 °C is important to reduce the series resistance of the LEDs due to blockage of acceptor-passivating hydrogen diffusion through the n-type doped top layer. Secondary ion mass spectroscopy measurements reveal Mg-incorporation into the topmost GaN:Ge layer, implying a non-abrupt p-n tunnel junction and increased depletion width. Still, significantly improved lateral current spreading as compared to conventional semi-transparent Ni/Au p-contact metallization and consequently a more homogeneous electroluminescence distribution across 1 × 1 mm2 LED structures is achieved. Direct estimation of the depletion width is obtained from electron holography experiments, which allows for a discussion of the possible tunneling mechanism.

Vacancy-type defects in Mg-doped GaN grown by ammonia-based molecular beam epitaxy probed using a monoenergetic positron beam

Vacancy-type defects in Mg-doped GaN were probed using a monoenergetic positron beam. GaN films with a thickness of 0.5–0.7 μm were grown on GaN/sapphire templates using ammonia-based molecular beam epitaxy and characterized by measuring Doppler broadening spectra. Although no vacancies were detected in samples with a Mg concentration [Mg] below 7 × 1019 cm−3, vacancy-type defects were introduced starting at above [Mg] = 1 × 1020 cm−3. The major defect species was identified as a complex between Ga vacancy (VGa) and multiple nitrogen vacancies (VNs). The introduction of vacancy complexes was found to correlate with a decrease in the net acceptor concentration, suggesting that the defect introduction is closely related to the carrier compensation. We also investigated Mg-doped GaN layers grown using In as the surfactant. The formation of vacancy complexes was suppressed in the subsurface region (≤80 nm). The observed depth distribution of defects was attributed to the thermal instability of the defects, which resulted in the introduction of vacancy complexes during the deposition process.

Atom probe tomography study of Mg-doped GaN layers

Atom probe tomography studies on highly Mg-doped homoepitaxial GaN (0001) layers with concentrations of 5 × 1019 cm−3 and 1 × 1020 cm−3 were performed. Mg cluster formation was observed only in the higher doped sample whereas in the lower doped sample the Mg distribution was homogeneous. CL measurements have shown that the emission normally attributed to stacking faults was only present in the lower doped layers (with Mg concentration of ∼5 × 1019 cm−3 or less), but absent in the higher doped layer, where Mg clusters were detected. Mg clusters are proposed to produce a screening effect, thereby destroying the exciton binding on the SFs and thus rendering them optically inactive.

High efficiency and enhanced ESD properties of UV LEDs by inserting p-GaN/p-AlGaN superlattice

Significantly improved electrostatic discharge (ESD) properties of InGaN/GaN-based UV light-emitting diode (LED) with inserting p-GaN/p-AlGaN superlattice (p-SLs) layers (instead of p-AlGaN single layer) between multiple quantum wells and Mg-doped GaN layer are reported. The pass yield of the LEDs increased from 73.53% to 93.81% under negative 2000 V ESD pulses. In addition, the light output power (LOP) and efficiency droop at high injection current were also improved. The mechanism of the enhanced ESD properties was then investigated. After excluding the effect of capacitance modulation, high-resolution X-ray diffraction (XRD) and atomic force microscope (AFM) measurements demonstrated that the dominant mechanism of the enhanced ESD properties is the material quality improved by p-SLs, which indicated less leakage paths, rather than the current spreading improved by p-SLs.