GaN and AlGaN have shown great potential in next-generation high-power electronic devices; however, they are plagued by a high density of interface states that affect device reliability and performance, resulting in large leakage current and current collapse. In this review, the authors summarize the current understanding of the gate leakage current and current collapse mechanisms, where awareness of the surface defects is the key to controlling and improving device performance. With this in mind, they present the current research on surface states on GaN and AlGaN and interface states on GaN and AlGaN-based heterostructures. Since GaNand AlGaN are polar materials, both are characterized by a large bound polarization charge on the order of 1013charges/cm2 that requires compensation. The key is therefore to control the compensation charge such that the electronic states do not serve as electron traps or affect device performance and reliability. Band alignment modeling and measurement can help to determine the electronic state configuration. In particular, band bending can determine how the polarization bound charge is compensated;however, the band bending is extremely sensitive to the specific processing steps such as cleaning, dielectric or metal deposition, postdeposition or postmetallization treatments, which affect oxygen coverage, carbon contamination,structural defects, bonding configurations, defect states, absorbates, and Fermi pinning states. In many cases, the specific effects of these treatments on the surface and interface states are not entirely clear as the nature of the electronic states has been obscured in complexity and subtlety. Consequently, a more systematic and methodical approach may be required.

This research focuses on the formation of Ag nanopatterns on periodically poled lithium niobate (PPLN). The photo-induced process employs UV-light exposure while the PPLN is immersed in a AgNO3 solution. The Ag deposition was consistent with previous results, showing preferential deposition along the domain boundary as well as an increased density of particles on the positive domain surface in comparison to the negative domain. By tuning the chemical solution concentration and the UV-light intensity, the Ag+ ion flux and the electron flux are varied and the deposition pattern could be controlled to either enhance the nanowire-like structures along the domain boundary or create a more uniform deposition pattern over the positive and negative domains. To understand the deposition process, we investigated the relationship between the Ag+ ion flux because of diffusion and the electron flux initiated by the UV exposure of the ferroelectric surface. The subsequent results suggest that this relationship is responsible for the different deposition patterns. The observed variation of boundary-enhanced or boundary-depressed deposition is explained by consideration of the electric field distribution and the ratio of the Ag+ ion and photon flux. The results establish that the ratio can be controlled by varying the solution concentration and/or UV-light intensity to generate enhanced nanowire-like structures along the domain boundary or a more uniform deposition pattern over the positive and negative surface. Moreover, the specific value of the Ag+/photon flux ratio where the pattern changes is dependent on other factors including the nucleation limited growth mechanism and the Stern layer on the lithium niobate.

Al2O3 films, HfO2 films, and HfO2/Al2O3 stacked structures were deposited on n-type, Ga-face, GaN wafers using plasma-enhanced atomic layer deposition (PEALD). The wafers were first treated with a wet-chemical clean to remove organics and an in-situ combined H2/N2 plasma at 650 °C to remove residual carbon contamination, resulting in a clean, oxygen-terminated surface. This cleaning process produced slightly upward band bending of 0.1 eV. Additional 650 °C annealing after plasma cleaning increased the upward band bending by 0.2 eV. After the initial clean, high-k oxide films were deposited using oxygen PEALD at 140 °C. The valence band and conduction band offsets (VBOs and CBOs) of the Al2O3/GaN and HfO2/GaN structures were deduced from in-situ x-ray and ultraviolet photoemission spectroscopy (XPS and UPS). The valence band offsets were determined to be 1.8 and 1.4 eV, while the deduced conduction band offsets were 1.3 and 1.0 eV, respectively. These values are compared with the theoretical calculations based on the electron affinity model and charge neutrality level model. Moreover, subsequent annealing had little effect on these offsets; however, the GaN band bending did change depending on the annealing and processing. An Al2O3 layer was investigated as an interfacial passivation layer (IPL), which, as results suggest, may lead to improved stability, performance, and reliability of HfO2/IPL/GaN structures. The VBOs were ∼0.1 and 1.3 eV, while the deduced CBOs were 0.6 and 1.1 eV for HfO2 with respect to Al2O3 and GaN, respectively.

The effects of surface pretreatment, dielectric growth, and post deposition annealing on interface electronic structure and polarization charge compensation of Ga-and N-face bulk GaN were investigated. The cleaning process consisted of an ex-situ wet chemical NH 4 OH treatment and an in-situ elevated temperature NH 3 plasma process to remove carbon contamination, reduce oxygen coverage, and potentially passivate N-vacancy related defects. After the cleaning process, carbon contamination decreased below the x-ray photoemission spectroscopy detection limit, and the oxygen coverage stabilized at $1 monolayer on both Ga-and N-face GaN. In addition, Ga-and N-face GaN had an upward band bending of 0.8 6 0.1 eV and 0.6 6 0.1 eV, respectively, which suggested the net charge of the surface states and polarization bound charge was similar on Ga-and N-face GaN. Furthermore, three dielectrics (HfO 2 , Al 2 O 3 , and SiO 2 ) were prepared by plasma-enhanced atomic layer deposition on Ga-or N-face GaN and annealed in N 2 ambient to investigate the effect of the polarization charge on the interface electronic structure and band offsets. The respective valence band offsets of HfO 2 , Al 2 O 3 , and SiO 2 with respect to Ga-and N-face GaN were 1.4 6 0.1, 2.0 6 0.1, and 3.2 6 0.1 eV, regardless of dielectric thickness. The corresponding conduction band offsets were 1.0 6 0.1, 1.3 6 0.1, and 2.3 6 0.1 eV, respectively. Experimental band offset results were consistent with theoretical calculations based on the charge neutrality level model. The trend of band offsets for dielectric/GaN interfaces was related to the band gap and/or the electronic part of the dielectric constant. The effect of polarization charge on band offset was apparently screened by the dielectric-GaN interface states. V C 2014 AIP Publishing LLC.

Variable intensity photoconductivity (PC) performed under vacuum at 325 nm was used to estimate drift mobility (μ) and density (σs) of negative surface charge for c-axis oriented Si-doped GaN nanowires (NWs). In this approach, we assumed that σs was responsible for the equilibrium surface band bending (∅) and surface depletion in the absence of illumination. The NWs were grown by molecular beam epitaxy to a length of approximately 10 μm and exhibited negligible taper. The free carrier concentration (N) was separately measured using Raman scattering which yielded N = (2.5 ± 0.3)  × 1017 cm−3 for the growth batch studied under 325 nm excitation. Saturation of the PC was interpreted as a flatband condition whereby ∅ was eliminated via the injection of photogenerated holes. Measurements of dark and saturated photocurrents, N, NW dimensions, and dimensional uncertainties, were used as input to a temperature-dependent cylindrical Poisson equation based model, yielding σs in the range of (3.5 to 7.5) × 1011 cm−2 and μ in the range of (850 to 2100) cm2/(V s) across the (75 to 194) nm span of individual NW diameters examined. Data illustrating the spectral dependence and polarization dependence of the PC are also presented. Back-gating these devices, and devices from other growth batches, as field effect transistors (FETs) was found to not be a reliable means to estimate transport parameters (e.g., μ and σs) due to long-term current drift. The current drift was ascribed to screening of the FET back gate by injected positive charge. We describe how these gate charging effects can be exploited as a means to hasten the otherwise long recovery time of NW devices used as photoconductive detectors. Additionally, we present data illustrating comparative drift effects under vacuum, room air, and dry air for both back-gated NW FETs and top-gated NW MESFETs.