Oxide-nitride Interface Research Articles

A junctionless dual-gate MOSFET-based programmable inverter for secured hardware applications using nitride charge trapping

In this paper we investigate a junction-less dual-gate metal–oxide–semiconductor field effect transistor (JL-DG-MOSFET)-based programmable inverter with an oxide–nitride–oxide (SiO2/Si3N4/SiO2) gate stack, which offers short-/long-term memory as well as logic functionalities depending on charge trapping in the nitride layer. It has been shown that the pulsing interval plays a pivotal role in deciding the short-term plasticity/long-term plasticity window based on the charges trapped/detrapped at/near the oxide–nitride interface. Moreover, we have demonstrated a JL-DG-MOSFET-based complementary metal–oxide–semiconducor inverter with a programmable switching threshold and propose a scheme for secure key generation for authentication. The intra-Hamming distance among the 21 keys generated by the programmable inverter is also depicted to demonstrate the efficacy of the proposed framework. This will eliminate the physical separation between the logic and memory and can offer attractive solutions for silicon-based low-power neuromorphic computing and hardware security.

Correlation between crystal structure and magnetism in PLD grown epitaxial films of ε-Fe2O3 on GaN

ABSTRACT In the present paper we discuss correlations between crystal structure and magnetic properties of epitaxial ε-Fe2O3 films grown on GaN. The large magnetocrystalline anisotropy and room temperature multiferroic properties of this exotic iron oxide polymorph, make it a perspective material for the development of low power consumption magnetic media storage devices. Extending our recent progress in PLD growth of ε-Fe2O3 on the surface of technologically important nitride semiconductors, we apply reciprocal space tomography by electron and x-ray diffraction to investigate the break of crystallographic symmetry occurring at the oxide-nitride interface resulting in the appearance of anisotropic crystallographic disorder in the sub-100 nm ε-Fe2O3 films. The orthorhombic-on-hexagonal nucleation scenario is shown responsible for the development of a peculiar columnar structure observed in ε-Fe2O3 by means of HRTEM and AFM. The complementary information on the direct and reciprocal space structure of the columnar ε-Fe2O3 films is obtained by various techniques and correlated to their magnetic properties. The peculiar temperature dependence of magnetization studied by the small-field magnetization derivative method and by neutron diffraction reveals the existence of a magnetic softening below 150 K, similar to the one observed earlier solely in nanoparticles. The magnetization reversal in ε-Fe2O3 films probed by X-ray magnetic circular dichroism is found different from the behavior of the bulk averaged magnetization measured by conventional magnetometry. The presented results fill the gap between the numerous studies performed on randomly oriented ε-Fe2O3 nanoparticles and much less frequent investigations of epitaxial epsilon ferrite films with lattice orientation fixed by the substrate.

Composition, structure and adhesion of cathodic-arc and glow-discharge deposited TiN/TiO2 coatings: a case study

Titanium-based vacuum coatings, such as TiN and TiO2, are being widely used for functionalization of the surfaces and for forming protective surface layers for a number of applications. In this study, TiN/TiO2 coatings were deposited by two physical vacuum deposition (PVD) methods: cathodic-arc (for TiN) and glow-discharge (for TiO2) deposition. We studied the bilayer film composition, structure and properties by means of X-ray diffraction (XRD) analysis, X-ray photoelectron spectroscopy (XPS), and scratch test analysis. The XRD spectrum indicated the presence of both anatase and rutile TiO2 phases, together with TiN reflections. The Ti 2p, O 1s and N 1s core level of XPS spectra in the depth of the oxide up to the oxide-nitride interface indicated intimate heterojunction between the layers. This results in a lower coefficient of friction and improved adhesion strength of the TiN/TiO2 film compared with pristine TiN coating.

Thermal stability of arc evaporated Al-Cr-O and Al-Cr-O/Al-Cr-N multilayer coatings

The thermal stability of cathodic arc evaporated Al0.70Cr0.30-based AlxCr1−xN/(AlxCr1−x)2O3 multilayer coatings and their monolithic constituents was investigated by combining differential scanning calorimetry and thermal gravimetric analysis with X-ray diffraction and transmission electron microscopy investigations.The formation of volatile Cr-oxides within the monolithic (Al,Cr)2O3 coating is suggested by a depletion of Cr during annealing treatments beginning with ~1050 °C. The structural integrity of the AlxCr1−xN/(AlxCr1−x)2O3 multilayer coatings — consisting of different interface architectures and number of layers — increases with an increasing oxide-nitride interface volume fraction up to ~1200 °C. The typically observed two-step dissociation process of AlxCr1−xN (to form Cr2N and finally Cr under N2-release) is reduced to a delayed one-step dissociation process.We could thus show, that combining AlxCr1−xN nitride coatings with (AlxCr1−x)2O3 is a promising approach in improving their thermal stability.

Experimental Extraction of the Charge Centroid of Holes Trapped in Metal–Oxide–Nitride–Oxide–Semiconductor Memories

In metal-oxide-nitride-oxide-silicon (MONOS)-type memories with an ultrathin tunnel oxide film, electrons and holes are injected from the silicon substrate to the silicon nitride charge trapping film owing to the quantum mechanical tunneling and are trapped in the film. In the previous studies, the constant-voltage carrier injection method has been used to determine the charge centroid of trapped carriers [1-4]. This method requires the application of an auxiliary pulse before injecting carriers. To determine the auxiliary pulse level, another set of measurements of the programming and erasing characteristics is needed in preliminary step. Such a sequence would generate an error in the charge centroid estimation due to the variation of the programming and erasing characteristics among individual memory elements. Therefore, we have developed the constant-current carrier injection (CCCI) method which does not require the application of the auxiliary pulse and any additional measurements [5]. The CCCI method is effective to obtain the accurate charge centroid of trapped carriers. On the other hand, these two methods can be used only in the case that the number of carriers passing through the silicon nitride film during the carrier injection is negligible as compared to that of carriers trapped in the nitride film. Therefore, neither of the two methods is usable with high gate voltages which induce a large number of carriers passing through the nitride film. In the present work, in order to obtain the charge centroid of holes trapped in the silicon nitride film at high gate voltages, we proposed a method based on the analysis of Fowler-Nordheim (F-N) tunneling current through the blocking oxide film. We determined the charge centroid and the density of holes trapped at high gate voltages by means of the proposed method and the CCCI method. Memory capacitors with the blocking oxide (SiOx)/silicon nitride/tunnel oxide stacked films were used in this work. A tunnel oxide film of 2.4 nm in thickness was grown by rapid thermal oxidation of silicon. A 30.4-nm-thick silicon nitride film was deposited at 600 °C in an LPCVD reactor. A blocking oxide film of 17.2 nm in thickness was grown at 400 °C using a PECVD method. The thickness of the blocking oxide film was designed to be sufficiently thick for electrons to obey the F-N tunneling mechanism and to gain a high kinetic energy in the oxide film. High-energy electrons entering the nitride film from the blocking oxide film would pass through the nitride film without interaction with nitride trap centers. Figure 1 shows the charge centroid of holes trapped in the nitride film, which was obtained by means of the analysis of F-N current under negative gate bias and the CCCI method. The charge centroid of trapped holes was initially located near the middle of the nitride film, and moved toward the blocking oxide-nitride interface as the flat-band voltage shift ΔVfb,h increased. Finally, the charge centroid reached 2.3 nm from the blocking oxide-nitride interface after the ΔVfb,h value was saturated. We also have estimated the density of trapped holes. A maximum of the density of trapped holes was determined to be 1.0×1013 holes/cm2 by means of the analysis of F-N tunneling current. The proposed method is useful to determine the maximum density of trapped holes which the charge trapping film in MONOS-type memories can achieve. Acknowledgment- We would like to express gratitude to S. R. A. Ahmed, K. Kato and H. Koizumi for their valuable discussions. This work was partly supported by JSPS KAKENHI Grant Number 26420280.

Origin of high temperature oxidation resistance of Ti–Al–Ta–N coatings

Alloying Ti–Al–N coatings with Ta has proven to enhance their hardness, thermal stability, and oxidation resistance. However, especially for arc-evaporated Ti–Al–Ta–N coatings only limited information on the detailed influence of the elements on various properties is available. Therefore, we have developed arc-evaporated Ti1−x−yAlxTayN coatings with various Al (x=0.50–0.65) and Ta (y=0.00–0.15) contents. While the thermal stability of our coatings during annealing in inert He atmosphere increases with increasing Ta content, best results are obtained for specific Ta–Al ratios during oxidation. Single phase cubic Ti0.32Al0.60Ta0.08N yields a mass-gain of only ~5% after 5h at 950°C in synthetic air, whereas Ti0.35Al0.65N is completely oxidized after 15min. This is in part based on the suppressed anatase and direct rutile TiO2 formation at a defined Ta–Al content. Consequently, the anatase-to-rutile transformation, generally observed for Ti1−xAlxN, is absent. This reduces the generation of pores and cracks within the oxide scale and especially at the nitride–oxide interface, leading to the formation of a protective rutile and corundum based oxide scale. This is also reflected in the pronounced decrease in activation energy for the protective scale formation from 232kJ/mol for Ti0.35Al0.65N down to 14.5kJ/mol for Ti0.32Al0.60Ta0.08N. Based on our results we can conclude that especially phase transformations within the oxide scale need to be suppressed, as the connected volume changes lead to the formation of cracks and pores.

Influence of Surface Microstructure and Chemical Composition on the Corrosion Resistance of Plain Steel Modified by Plasma-Assisted Diffusion

The influence of surface microstructure and chemical composition on the corrosion behavior of AISI 1045 (UNS G10450) plain steel modified by plasma-assisted diffusion of nitrogen and oxygen was investigated. A detailed surface characterization was performed before and after cyclic potentiodynamic polarization and salt spray tests, using scanning electron microscopy at low and high resolutions, energy-dispersive x-ray spectroscopy, x-ray diffraction, and glow discharge optical emission spectroscopy. The corrosion resistance as determined by cyclic potentiodynamic polarization was found to depend more strongly on the morphology and composition of the outermost oxynitride layer than on its thickness. For post-oxidation times below and above 90 min, the oxynitride layer presented defects that can act as pathways for corrosive species. The degradation of the corrosion resistance for longer processing times is through cracking of granules, which were previously formed at shorter times owing to the hydrogen accumulation at the oxide-nitride interface. The salt spray tests revealed a main dependence on the manganese sulfide (MnS) inclusions content in the base material.

Oxidation and diffusion study on AlCrVN hard coatings using oxygen isotopes 16O and 18O

The present work aims to investigate the oxidation behaviour of AlCrVN hard coatings of equal composition but of different crystal structure. In order to gain more information about the mechanisms that are active during oxidation, a two-stage oxidation procedure has been applied where different isotopes, 16O and 18O, were introduced in each step. The analysis by means of secondary ion mass spectrometry depth profiling with its inherent isotope selectivity provided information on the general oxidation behaviour as well as the oxygen diffusion during the oxidation process. The single-phase coating with its face-centred cubic (fcc) structure presents a higher oxidation resistance as compared to the dual-phase coating containing a wurtzite and an fcc phase. After the annealing treatment the surface of the latter is entirely covered by VO 2 and V 2O 5 as evidenced by Raman spectroscopy. The single-phase coating, on the other hand, reveals unoxidised coating material and AlVO 4 crystals. However, even though exhibiting a significantly different oxidation resistance, the oxygen diffusion is similar. In both cases the peak values of 18O, which was introduced in the second stage, were found near the oxide–nitride interface indicating that O atoms diffused through the already formed oxides. Additional experiments using a gas mixture comprising natural water vapour H 2 16O and 18O 2 revealed that mainly the presence of molecular oxygen causes oxidation as with increasing water vapour partial pressure the oxide layer thickness was significantly reduced.

Memory effect of a single-walled carbon nanotube on nitride-oxide structure under various bias conditions

We report on the memory effect of single-walled carbon nanotubes (SWNTs) placed on a nitride-oxide layer structure designed as a charge storage medium. The conductance of the SWNT was modulated by the injected charge in the nitride-oxide interface and the polarities of injected charges were then detected. A large on/off-state current ratio (>104) was obtained at a small program/erase voltage range (<3 V). We also studied the effect of a half-selected cell on the conductance of the SWNTs to identify the issues with cross-point memory architecture.

On the effect of Ta on improved oxidation resistance of Ti–Al–Ta–N coatings

Formation of protective oxide scales is the main reason for the high oxidation resistance of TiAlN based coatings. Here the authors report on further improvement in the oxidation resistance of TiAlN by Ta alloying. An industrial-scale cathodic arc evaporation facility was used to deposit Ti–Al–Ta–N coatings from powder metallurgically produced Ti38Al57Ta5 targets. After oxidation in ambient air, a significantly reduced oxide layer thickness in comparison to unalloyed TiAlN reference material was observed. Energy-dispersive x-ray spectroscopy line scans and secondary ion mass spectroscopy depth profiling showed that the oxide scale consists of an Al-rich top layer without detectable amount of Ta and a Ti–Ta-rich sublayer. Transmission electron microscopy investigations revealed α-Al2O3, rutile-type TiO2, and anatase-type TiO2 as the scale forming oxides. Furthermore, the Ti–Ta-rich sublayer consists of a porous layer at the oxide-nitride interface but appears dense toward the Al-rich top layer. The improved oxidation resistance is explained by doping the TiO2 lattice by replacing Ti4+ with Ta5+ in the rutile lattice, which decreases the oxygen mass transport. This leads to reduced oxidation of Ti under formation of TiO2 at the oxide-nitride interface and is the reason for the excellent oxidation behavior of Ti–Al–Ta–N coatings.

Charge trapping properties at silicon nitride/silicon oxide interface studied by variable-temperature electrostatic force microscopy

Charge trapping properties of electrons and holes in ultrathin nitride-oxide-silicon (NOS) structures were quantitatively determined by variable-temperature electrostatic force microscopy (EFM). From charge retention characteristics obtained at temperatures between 250 and 370°C and assuming that the dominant charge decay mechanism is thermal emission followed by oxide tunneling, we find that there are considerable deep trap centers at the nitride-oxide interface. For electron, the interface trap energy and density were determined to be about 1.52eV and 1.46×1012cm−2, respectively. For hole, these are about 1.01eV and 1.08×1012cm−2, respectively. In addition, the capture cross section of electron can be extracted to be 4.8×10−16cm2. The qualitative and quantitative determination of charge trapping properties and possible charge decay mechanism reported in this work can be very useful for the characterization of oxide-nitride-silicon based charge storage devices.

A model of radiation effects in nitride–oxide films for power MOSFET applications

Total dose effects in composite nitride–oxide (NO) films with various nitride and oxide thicknesses are studied. Metal nitride oxide semiconductor (MNOS) devices with thicknesses in the range of power transistors have been irradiated with X-rays up to a total dose of 1 Mrad (SiO 2). A model is proposed to describe the observed flatband voltage shifts as a function of total dose in the range of electric fields where the charge yield in the dielectrics to a first order approximation is linearly proportional to the electric field. The modulation of electric field in the dielectrics due to charge accumulation at the NO interface is found to be the dominating factor in determining the flatband voltage shift. It is suggested that, for the process conditions examined here, n-channel MNOS devices with oxide thickness between 12 and 15 nm give minimal radiation-induced flatband voltage shifts.

Effect of polysilicon gate type on the flatband voltage shift for ultrathin oxide-nitride gate stacks

In this work, we demonstrate that the magnitude of flatband voltage (V/sub FB/) shift for ultrathin (<2 nm) silicon dioxide-silicon nitride (ON) gate stacks in MOSFET's depends on the Fermi level position in the gate material. In addition, a fixed positive charge at the oxide-nitride interface was observed.

The initial oxidation of ε-Fe 2N 1− x: an XPS investigation

The initial oxidation of α-Fe and ε-Fe 2N 1− x , subjected either to a sputter cleaning pretreatment or a sputter cleaning plus additional annealing pretreatment, was investigated with XPS. The samples were oxidised at p O 2 =8·10 −5 Pa and temperatures ranging from 300 to 600 K. From the Fe 2p and O 1s spectra the thickness and composition of the oxide film was determined. The composition of the oxide films formed on α-Fe and on ε-Fe 2N 1− x was only a function of oxidation temperature and film thickness and was independent of the composition or pretreatment of the substrate. Analysis of the N 1s spectra provided information on the electric charge on nitrogen atoms and the depth distribution of nitrogen. A linear relation was found between the N 1s electron binding energy and the nitrogen concentration in the substrate. Upon oxidation of the iron nitride, nitrogen atoms accumulated underneath the oxide film. If the nitrogen concentration at that location exceeded the maximum solubility of nitrogen in ε-Fe 2N 1− x an additional N peak appeared in the N 1s spectrum, which indicated the formation of a nitrogen containing phase other than ε-Fe 2N 1− x at the nitride–oxide interface. For the oxygen exposures applied, the oxide-film thickness decreased with increasing nitrogen concentration in the substrate. The effect of nitrogen in the substrate on the initial oxidation was evaluated from the results.

Lateral resolution of light-addressable potentiometric sensors: an experimental and theoretical investigation

The surface potential of semiconductor devices in contact with electrolyte solutions is an important part of signal transduction for a variety of bioanalytical devices. Here we have investigated the lateral resolution at which the surface potential may be measured with a semiconductor-based device, a light-addressable potentiometric sensor (LAPS). We have first established an experimental setup where a permanent charge pattern is generated in the oxide-nitride interface of an n-doped silicon wafer by UV irradiation. Using a laser beam to interrogate the LAPS, the charge pattern can be detected by measuring the local photocurrent at a resolution of better than 100 μm. A theoretical model based on the diffusion and recombination of photogenerated minority charge carriers has been developed and solved analytically; it is consistent with experiment. For Beer-Lambert law absorption of a sufficiently narrow beam of interrogating light, according to the theory the lateral resolution depends on the relative sizes of the penetration depth of the light, d, and the recombination-diffusion length of the carriers, L. When d ⪡ L, the resolution is (2Ld) 1 2 ; when L ⪢ d, it is (2L 2) 1 2 .

Mechanical properties and oxidation behaviour of HIPed hafnium nitride ceramics

Abstract The compaction at 1600°C, under an isostatic argon pressure of 195 MPa, of a coarse hafnium nitride powder, previously encapsulated in a vacuumed titanium can, has led to the formation of a porous material (dr ≅ 85%). In order to produce a high-performance material, the same HfN powder has been encapsulated into a silica container and hot isostatically pressed at 1950°C, under a pressure of 195 MPa. In both cases, the dwell time lasted 1 h. Under these conditions, the dense material (dr > 95%) had a Vickers microhardness Hv10N of 1450, a modulus of rupture rather low (350 MPa) and a SENB fracture toughness of 4·5 MPa m The oxidation behaviour, in dry oxygen, suggested that the reaction, between 755 and 950°C, was governed by an interfacial process occurring at the nitride-oxide interface, similarly to the behaviours of dense HIPed TaN and ZrN materials.

Carrier transport and storage in Si 3N 4 for metal-nitride-oxide-semiconductor memory applications

The carrier transport and storage properties in the nitride of metal-nitride-oxide-semiconductor (MNOS) capacitors have been studied by combining repeated voltage ramp stress and constant voltage stress measurements. From the positive flat band voltage shift that results after bias application to the samples, it is clear that electrons dominate in the transport current and trapping. However, it is shown that, whereas at intermediate and low fields electrons are the single species involved in the capture-emission processes in the nitride, at higher electric fields a certain mechanism responsible for positive charge accumulation in the oxide-nitride interface region is also evident. By fitting theoretical results describing electron trapping in the nitride to the experimental flat band voltage shift dependence on bias, the density of electron traps in the nitride and the trapping efficiency have been estimated to be 3.7 × 10 18 cm −3 and 10 −2s cm 2 respectively. We have also compared the experimental flat band voltage shift dependence on current obtained in constant current stressed MNOS capacitors with theoretical values derived from the fitted parameters. Quite good agreement has been found (except at low stress levels where leakage currents through the parallel resistance of the current source become significant), thus supporting the validity of the proposed model.

Transformation of silicon nitride in oxygen plasma

The oxygen plasma treatment of chemical-vapour-deposited Si 3N 4 thin films at floating potential and low T (less than 600 °C) leads to the growth of a silicon oxide surface layer and to a decrease of the nitrogen content of the samples. Nuclear reaction analysis was used to study the composition of the films, the oxide growth kinetics and the loss of nitrogen atoms. After an initial transient oxidation regime, parabolic growth laws were found, indicating that the rate-limiting step in the oxide formation is a diffusion process and not a surface reaction. In the low temperature regime the oxide growth rate and the activation energy were found to be similar to those obtained during the plasma oxidation of bare silicon substrates. 18O-enriched plasmas were used to study the nitride transformation, and the 18O depth distribution was determined by resonance depth profiling. The 18O atoms were found at the surface of the sample, forming a pure SiO 2 layer on the underlying Si 3N 4. The absence of a large mixing at the oxide-nitride interface indicates that the oxide growth occurs by a layer-by-layer oxidation process and not by the formation of an oxynitride compound of varying composition with depth.

Physical origin of long-term charge loss in floating-gate EPROM with an interpoly oxide-nitride-oxide stacked dielectric

The long-term charge loss characteristic in a floating-gate EPROM cell with an oxide-nitride-oxide (ONO) interpoly stacked dielectric has been studied quantitatively. It is found that trapped electrons at the nitride-oxide interface can directly tunnel through a thin approximately 30-AA top oxide. The estimated tunneling barrier height is about 2.6 eV, which is consistent with the previous result based on MNOS studies. The thermal activation of the long-term charge loss is believed to be caused by the availability of trapped electrons at the top oxide-nitride interface. The thermal activation energy of the long-term charge loss is about 0.37 eV, which is similar to the activation energy of the second phase. It is believed that the trapped electron density at the top oxide-nitride interface becomes larger at higher temperatures due to the activated electron movement during the second phase. >

Electron and hole charge separation with a dual channel transistor

Summary form only given. The authors have fabricated a n/sup +/ poly gate dual channel memory transistor with a 0.75- mu m-deep n-buried layer on p-Si, and a triple dielectric (ONO) sandwich comprising a 20- mu A tunneling oxide, 115 AA of LPCVD (low-pressure chemical vapor deposition) deposited Si/sub 3/N/sub 4/, and 52 AA of top oxide. For positive gate bias beyond flat band, electrons are collected at the substrate. For negative gate bias beyond flat band, holes are supplied by the surface channel contacts and the electrons, spatially separated from the holes by the potential well in the buried channel, are collected at the buried-channel contacts. The data for electron and hole injection into the nitride show centroids (from the tunneling oxide-nitride interface) to be approximately equal to the nitride thickness for small level injection, decreasing with increasing injection and saturation beyond 1 mu C/cm/sup 2/. The saturation values for electron and hole centroids are 75+or-5 AA and 55+or-2 AA, respectively. >