Disorder Induced Gap State Model Research Articles

Effects of surface plasma treatment on threshold voltage hysteresis and instability in metal-insulator-semiconductor (MIS) AlGaN/GaN heterostructure HEMTs

In a bid to understand the commonly observed hysteresis in the threshold voltage (VTH) in AlGaN/GaN metal-insulator-semiconductor high electron mobility transistors during forward gate bias stress, we have analyzed a series of measurements on devices with no surface treatment and with two different plasma treatments before the in-situ Al2O3 deposition. The observed changes between samples were quasi-equilibrium VTH, forward bias related VTH hysteresis, and electrical response to reverse bias stress. To explain these effects, a disorder induced gap state model, combined with a discrete level donor, at the dielectric/semiconductor interface was employed. Technology Computer-Aided Design modeling demonstrated the possible differences in the interface state distributions that could give a consistent explanation for the observations.

On the origin of interface states at oxide/III-nitride heterojunction interfaces

The energy spectrum of interface state density, Dit(E), was determined at oxide/III-N heterojunction interfaces in the entire band gap, using two complementary photo-electric methods: (i) photo-assisted capacitance-voltage technique for the states distributed near the midgap and the conduction band (CB) and (ii) light intensity dependent photo-capacitance method for the states close to the valence band (VB). In addition, the Auger electron spectroscopy profiling was applied for the characterization of chemical composition of the interface region with the emphasis on carbon impurities, which can be responsible for the interface state creation. The studies were performed for the AlGaN/GaN metal-insulator-semiconductor heterostructures (MISH) with Al2O3 and SiO2 dielectric films and AlxGa1–x layers with x varying from 0.15 to 0.4 as well as for an Al2O3/InAlN/GaN MISH structure. For all structures, it was found that: (i) Dit(E) is an U-shaped continuum increasing from the midgap towards the CB and VB edges and (ii) interface states near the VB exhibit donor-like character. Furthermore, Dit(E) for SiO2/AlxGa1–x/GaN structures increased with rising x. It was also revealed that carbon impurities are not present in the oxide/III-N interface region, which indicates that probably the interface states are not related to carbon, as previously reported. Finally, it was proven that the obtained Dit(E) spectrum can be well fitted using a formula predicted by the disorder induced gap state model. This is an indication that the interface states at oxide/III-N interfaces can originate from the structural disorder of the interfacial region. Furthermore, at the oxide/barrier interface we revealed the presence of the positive fixed charge (QF) which is not related to Dit(E) and which almost compensates the negative polarization charge (Qpol−).

On the origin of the two-dimensional electron gas at AlGaN/GaN heterojunctions and its influence on recessed-gate metal-insulator-semiconductor high electron mobility transistors

It is commonly accepted that interface states at the passivation surface of AlGaN/GaN heterostructures play an important role in the formation of the 2DEG density. Several interface state models are cited throughout literature, some with discrete levels, others with different kinds of distributions, or a combination of both. The purpose of this article is to compare the existing interface state models with both direct and indirect measurements of these interface states from literature (e.g., through the hysteresis of transfer characteristics of Metal-Insulator-Semiconductor High Electron Mobility Transistors (MISHEMTs) employing such an interface in the gate region) and Technology Computer Aided Design (TCAD) simulations of 2DEG densities as a function of the AlGaN thickness. The discrepancies between those measurements and TCAD simulations (also those commonly found in literature) are discussed. Then, an alternative model inspired by the Disorder Induced Gap State model for compound semiconductors is proposed. It is shown that defining a deep border trap inside the insulator can solve these discrepancies and that this alternative model can explain the origin of the two dimensional electron gas in combination with a high-quality interface that, by definition, has a low interface state density.

Accumulation capacitance frequency dispersion of III-V metal-insulator-semiconductor devices due to disorder induced gap states

The origin of the anomalous frequency dispersion in accumulation capacitance of metal-insulator-semiconductor devices on InGaAs and InP substrates is investigated using modeling, electrical characterization, and chemical characterization. A comparison of the border trap model and the disorder induced gap state model for frequency dispersion is performed. The fitting of both models to experimental data indicate that the defects responsible for the measured dispersion are within approximately 0.8 nm of the surface of the crystalline semiconductor. The correlation between the spectroscopically detected bonding states at the dielectric/III-V interface, the interfacial defect density determined using capacitance-voltage, and modeled capacitance-voltage response strongly suggests that these defects are associated with the disruption of the III-V atomic bonding and not border traps associated with bonding defects within the high-k dielectric.

Rigorous analysis of the electronic properties of InP interfaces for gas sensing

A theoretical analysis of the surface Fermi level position ( E FS) at n-InP interfaces with different surface state density ( N SS( E)) and surface fixed charge ( Q FC) representing adsorbed ions or surface δ-doping has been performed in order to understand the InP-based gas sensor behaviour. Furthermore, the in-depth profiles of the potential barrier in equilibrium and under illumination (surface photovoltage) have been rigorously calculated. A U-shaped interface state continuum has been assumed in accordance with the Disorder Induced Gap State model. From the simulated dependencies of E FS vs. the minimum surface state density N SS0, the movement of E FS in the energy band gap as well as its pinning position have been investigated. In addition, the analysis of the E FS sensitivity to the negative and positive Q FC has revealed the remarkable charge detection sensitivity of InP interfaces within different dynamic ranges.

Scanning tunneling microscopy and x-ray photoelectron spectroscopy studies of atomic level structure and Fermi level pinning on GaAs(110) surfaces grown by molecular beam epitaxy

The relationship between atomic level structure and Fermi level pinning on n-type GaAs(110) surface grown by molecular beam epitaxy (MBE) is studied in situ by scanning tunneling microscopy (STM)/scanning tunnel spectroscopy (STS) and x-ray photoelectron spectroscopy (XPS). Fermi level by XPS on our MBE-grown n-GaAs(110) surfaces is found to lie at around EC−0.5 eV, indicating the presence of band bending. Among a few characteristic features observed by STM in addition to the basic (1×1) relaxation structure, the dominant holelike structure in the empty state image is correlated with band bending. It shows a gradual decrease of the apparent height over the several atomic distances toward the center in the empty state STM image and gradual increase of the apparent height in the filled state image, indicating the presence of negative charge. It accompanies random site deviations of surrounding atoms. The density of the holelike feature agrees roughly with the required number density of surface pinning centers to cause the observed band bending. This holelike structure originates most likely from a Ga vacancy. However, in STS spectroscopy measurements, a U-shaped continuous gap state distribution extending from the band edges, rather than a peaked discrete deep acceptor state distribution, is observed with a local downward shift of the surface Fermi level. A picture is presented where generation of Ga vacancy introduces local bond disorder and acts like a pinning center having a gap state continuum in accordance with a disorder-induced gap state model.

Evolution mechanism of nearly pinning-free platinum/n-type indium phosphide interface with a high Schottky barrier height by in situ electrochemical process

Recently, unusually high Schottky barrier heights (SBHs) have been realized by our group on n-type InP and related materials by an in situ electrochemical deposition of Pt. In an attempt to understand the underlying mechanism of the SBH enhancement, this article investigates in detail the evolution process of the metal (Pt, Ni, Co, and Ag)-InP interface during the in situ electrochemical process, using current–voltage, capacitance voltage, deep level transient spectroscopy, x-ray photoelectron spectroscopy, Raman, atomic force microscope, and scanning electron microscope measurements. Pt deposition by the electrochemical process realized an oxide-free, defect-free, stress-free, and nearly pinning-free interface, whereas Pt deposition by conventional electron beam evaporation and sputtering processes as well as Ag, Ni, and Co deposition by the electrochemical process gave rise to stressed and pinned interfaces. The observed large process dependence of SBH can be explained by none of the metal induced gap state model, the unified defect model, and the effective workfunction model. It is explained here by the disorder induced gap state model.

Characterization of InGaAs surface passivation structure having an ultrathin Si interface control layer

A recently proposed passivation structure of In0.53Ga0.47As surface is characterized by x-ray photoelectron spectroscopy (XPS), capacitance–voltage, and deep-level transient spectroscopy measurements. It uses an ultrathin molecular beam epitaxy Si interface control layer (ICL) and a photochemical vapor deposition (CVD) SiO2 outer insulator. XPS study indicates that growth of Si ICL is coherent and two-dimensional up to about 10 Å beyond which clustering is initiated, and that it is effective in preventing the InGaAs layer from selective oxidation of Ga during photo-CVD SiO2 deposition. C–V and deep level transient spectroscopy (DLTS) measurements give Nss distributions of very different shapes, the latter exhibiting two distinct peaks, which may be erroneously interpreted as defect levels. Suppression of selective oxidation of InGaAs does not alter the shapes of these distributions, although it reduces the magnitude. The results are explained by the disorder-induced gap state model rather than the defect model or the effective work function model.

A self-consistent computer simulation of compound semiconductor metal-insulator-semiconductor C-V curves based on the disorder-induced gap-state model

Compound semiconductor metal-insulator-semiconductor (MIS) capacitance-voltage (C-V) curves are simulated on a computer, assuming the presence of a disorder-induced gap-state (DIGS) continuum near the interface in which bonding and antibonding states are distributed both in space and in energy around a particular charge neutrality point, EHO . A program based on Shockley–Read–Hall statistics is developed and applied to simulate a set of experimental MIS C-V curves measured on Al2 O3 /native oxide/InP MIS capacitors. By assuming a particular type of energy and spatial distribution of DIGS continuum and a suitable single set of dynamic parameters, the simulation can completely and self-consistently reproduce the experimental MIS C-V curves with complex hysteresis behavior, which varies with the bias amplitude and swing speed. The excellent agreement between theory and experiment supports the DIGS model. The danger involved in simply applying Terman’s method to determine an Nss distribution is pointed out. A quick procedure for reasonably accurate Nss measurement is suggested.

Unified disorder induced gap state model for insulator–semiconductor and metal–semiconductor interfaces

The energy location for the interface state density Nss minimum of the insulator–semiconductor (I–S) interface and the Fermi-level pinning position at the metal–semiconductor (M–S) interface are shown to coincide and to lie at the same position of 5.0 eV from the vacuum level for major tetrahedral semiconductors. Neither the unified defect model nor the metal induced gap state model can explain the novel striking correlation between the I–S and M–S interfaces. The correlation as well as the observed peculiar photoionization behavior of the I–S interface are explained by the novel unified disorder induced gap state (DIGS) model where DIGS pin or restrict the movement of the surface Fermi level. The above characteristic energy, EHO, is shown to be the Fermi energy of the DIGS spectrum which is given by the hybrid orbital energy of the sp3 bond of the host. The DIGS model explains remarkably well the behavior of the M–S interface formed on the bare or oxide covered semiconductor surface as well as the various features of Nss distribution of the I–S interface. The correlation between the DIGS-free heterojunction (S–S) interface and M–S/I–S interface is explained by the fact that EHO is a universal reference energy level of the host which is invariant under any off-diagonal interactions, as is evidenced by the alignment of transition metal deep levels, DX centers and EL2 with respect to EHO. Band offset at the S–S interface is proposed to be determined by the alignment of EHO which inevitably involves formation of interface dipole when two EHO levels lie at different positions from the vacuum level.