The chemical stability of magneto-optic structures of sputter-deposited SiO, SiO2, or Si3N4/TbFeCo/SiO, SiO2, or Si3N4 was studied in detail. The degradation of the magnetic properties of the TbFeCo film have been correlated with the oxygen content within the film as determined by Auger and RBS. Auger and XPS depth profiles have revealed that for oxygen containing dielectrics the preferential oxidation of terbium is responsible for the observed differences in coercivity. Reduction of the dielectric at the SiO/TbFeCo interface, along with diffusion of oxygen into the RETM film, have been observed. The Si3N4/TbFeCo structure shows no evidence of either interdiffusion between the layers or oxidation of the TbFeCo film. From our investigation we conclude that Si3N4 is a superior dielectric for quadri layer structures in magneto-optic media.

Silicon dioxide layers were grown on single crystal (100) β-SiC between 1000°C and 1200°C, in wet O2, dry O2, and wet Ar. All processes demonstrated a linear-parabolic relationship with time. Both wet processes had a slower rate than dry oxidation at 1050°C and below. The activation energies of the linear and parabolic rate constants were calculated for each process. The activation energy of the parabolic rate constant for wet oxidation was found to be inversely dependent on the amount of oxygen present as a carrier gas.

The dry oxides exhibited a very flat surface; in contrast, SEM and XTEM reveal that wet oxidation preferentially oxidizes dislocation bands, causing raised lines on the oxide and corresponding grooves in the SiC. It is proposed that the much higher solubility of H2O in SiO2 as compared with that of O2 (103 at 1000°C), allows wet oxidation to be preferential.

The electrical properties of the SiO2/SiC interface were studied using the conductance vs voltage (G-V) data for the metal-oxide-SiC (MOS) structure. It was found that the dry oxide contained too mjch charge either at the oxide-SiC interface or within the oxide films to obtain useful data. On the other hand the wet oxide invariably resulted in better capacitance and conductance data. The capacitance-voltage data showed that the SiC surface exhibited accumulation, depletion or inversion when the appropriate gate bias was applied. The conductance-voltage data indicate electronic surface states at the oxide-SiC interface. From this conductance data the interface state density has been estimated.

Thermal oxidation of high dose arsenic implanted p-type <100> silicon has been performed in order to study arsenic snowplowing. Bubble pattern formation and local oxide bowing (mound) in relation to the arsenic concentration in silicon as well as an HC1 ambient have been observed for an oxidation temperature of 1050°C. Oxide peel off at an oxidation temperature of 950°C for an atmosphere with an HC1 ambient has been observed. In addition, an anomaly in the temperature dependence of the oxidation rate independent of an HC1 ambient has been found.

Both thin, 60→140Å thick, and native oxides on As and P doped Si were studied via null ellipsometry and angle resolved ESCA. Results showed higher oxide growth rates to occur for higher surface concentrations of a given dopant. For similar surface concentrations, P-doped samples oxidized faster than As-doped, in both cases. Exact oxidation rates were not determined as results indicate that stoichiometric variations and oxidation of carbon contamination influence thickness determination of these oxides.

Constant-voltage stressing has been used to investigate the damage of SiO2 and the Si/SiO2 interface induced by silicon implantation through polysilicon/SiO2/P-Si structures annealed at 950°C. The implant doses used were from 5×1011cm-2 to 1014cm-2. Although no detectable interface states density was observed after the annealing for implant doses less than 2×1013cm-2, interface states generation, hole trapping, and electron trapping were found to be greatly enhanced by the Si implantation. The interface states density generation rate was found to increase with higher implant doses. The density of hole trapping centers saturated at a value of 3×1012cm-2 for implant doses higher than 2×1012cm∼-2. The density of electron trapping centers was found to increase with implant dose, while the associated trapping cross section was much smaller than that of the unimplant oxide.

We have used the DLTS (Deep Level Transient Spectroscopy) technique to characterize the traps of the oxide layer near the Si-SiO2 interface. Using the so-called saturating pulse DLTS, no saturation is observed when all the fast interface states are filled. This unusual behavior is interpreted by taking into account tunneling emission from oxide traps situated near the Si-SiO2 interface. We show how the deconvolution between interface states and oxide traps is allowed by introducing a model of tunneling emission from oxide traps into Si substrate, for the derivation of the DLTS response. By this way a depth localisation of these oxide traps is obtained : about 10 to 30 Å from the interface. The experimental study of the filling kinetics allows the determination of the concentration of these oxide traps (≈ 107 to 108 cm-2) and their associated tunneling capture cross-section (≈ 10-22 to 10-21 cm2). The validity of the model and measurements is rapidly discussed.

The phases, morphologies and microstructures of native oxides and nitrides, grown on the vicinal planes of germanium, are discussed. Thermal oxides, formed under high pressure, were shown to be primarily amorphous for (100) and (110) oriented substrates and intermixed with a crystalline hexagonal phase on the (111) surfaces. Thermal treatments, in one atmosphere of flowing ammonia gas, converted oxide films to mixtures of nitrides and oxynitrides with the nitrides found to be combinations of a- and β-Ge3N4. The α-phase formed from condensation of vapors above the surface whereas the β-phase was a solid-solid reaction product which initiates at the oxide/germanium interface. These two processes appeared to proceed independently of each other. Results of low angle X-ray diffraction (XRD), far infrared transmission (FIRT), scanning transmission electron microscopy (STEM) and X-ray photoelectron spectroscopy (XPS) are discussed.

We have investigated silicon oxide formation on co-deposited Au-Si alloy films at low temperatures. Alloy compositions spanning the Au-Si binary phase diagram have been examined. It was found that alloys with compositions above 40 at. % Au were most efficient in promoting oxide formation. These results are compared with the Au on Si sample configuration where Si is transported through the Au before oxidizing. The possibility of using alloys to form Au lines with a self passivating oxide coating after patterning and oxidation is discussed.

The differences of the room temperature oxidation behavior of ordered Ag/Si(111) and Au/Si(111) surfaces were studied by surface sensitive soft x-ray photoemission spectroscopy obtained with synchrotron radiation. Si surfaces covered with a monolayer of Ag or Au, once annealed to display a √3×√3 LEED pattern, were believed to be passivated against oxidation according to earlier reports. This work shows that these two surfaces oxidize but in a different way. Up to 104 L O2 exposures, the observed valence band of the Au/Si surface's valence band electron energy distribution curve is almost identical to that of the surface before oxygen exposure. But the corresponding Si 2p core level spectrum shows a small chemically shifted component indicating an initial stage of the formation of Si oxide. Thi3 chemically shifted signal becomes a strong peak at -3.7 eV below the clean Si position, characteristic of SiO2, after subsequent O2 exposures up to 1010 L. The Ag/Si system behaves in a similar fashion, but oxide growth saturates at 108 L, and the final oxides formed include a distribution of suboxides in addition to SiO2. Clearly, oxide formation is not prohibited by the presence of the ordered Au or Ag metal overlayer but delayed. Although the onset of oxidation is delayed compared to that for the clean Si surface, due to the metal-silicon bonding, the oxide formation is much faster once the surface starts to oxidize.

The reproducible growth of thin high quality silicon dioxide films is of fundamental importance to advanced VLSI processes. Oxides in the range 1–2 nm are required for electron tunnelling barriers in advanced BIPOLAR processes, and in the range 10–50 nm for the gate oxide in MOS processes.

The excellent degree of control over time, temperature and ambient atmosphere that can be achieved in halogen lamp rapid annealing systems suggested the possibility of growing thin films of silicon dioxide on silicon very reproducibly.

This paper compares the growth of oxides in an AG Associates' Heatpulse system with those grown in a conventional dry oxidation system. Oxides in the range 1–1000 nm have been grown in these systems and the thicknesses are compared with those predicted by the SUPREM 3 and Deal Grove models. Modifications to these models are suggested which allow for accurate prediction of thermal oxide growth above 2 nm.

A systematic study was performed to investigate the ion beam induced metal nitride formation by direct nitrogen implantation into 10 selected metals, either of thin film or bulk material. An X-ray diffractometer, together with a specially-designed Seemann-Bohlin attachment, was employed to provide fast and reliable phase identification in the thin implanted layer. The results show that room temperature nitrogen implantation can lead to the formation of many, but not all, of the equilibrium metal nitrides. The formation of metal nitrides by implantation is discussed in terms of the thermodynamic condition and the readiness of structural trans format ion.

The interfacial reactions and microstructures of metal-GaAs contacts are, in general, much more complicated and difficult to control than the corresponding metal-Si contacts. They are very sensitive to reaction temperature, ambient, metal layer thickness, and the GaAs surface cleaning procedure. Many of the ohmic and gate contacts to GaAs currently in use or under development for GaAs FET devices are comprised of more than one metal species and in some cases a doping element as well. In dealing with these complexities, information about the microstructure at the contact interface is critically needed: for the evaluation of a specific contact metallurgy, for the definition of an optimum fabrication process, and, most important of all, for generating new ideas for better contact schemes. In this paper, our TEM and STEM studies of several ohmic and gate contacts that are of technological interest will be described. Attention will be drawn to the link between the interfacial microstructures and their electrical behavior, the kinetics of interfacial reactions, and thermal stability. The current constraints on obtaining ideal, reliable and controllable metal-GaAs contacts will also be discussed.

The recent development of Pulsed Laser Atom Probe (PLAP) analysis has allowed routine analysis of the composition of a wide range of semiconducting materials. This paper presents results on the analysis of stoi chiometry variations in MOCVü GaAlAs layers demonstrating that accurate analysis of aluminium concentration fluctuations can be achieved with this technique.

A method for extracting quantitative information from weak diffraction features at or near matrix reflections is described. The technique requires an accurate knowledge of the fall-off of the matrix reflection. This fall-off was studied both experimentally and theoretically for a wide range of coherence conditions in the microscope. Subtraction of this matrix contribution from the total intensity of an [004] reflection from a E=13 (001) twist boundary shows that the streak arises from local as well as from long range interference effects. The latter contribution varies with spatial coherence length which depends on the condenser setting of the microscope.

The atomic structure of etched and polished Si (111) surfaces has been determined quantitatively on an atomic scale by SPA-LEED (spot profile analysis of low energy electron diffraction) in an ultrahigh vacuum. The energy dependant broadening of the diffracted electron beam yields the average size of step free terraces and the distribution of the atomic steps.

Whereas acid etched Si (111) wafers exhibit surfaces with very high step atom densities, alkaline etched Si (111) wafers show atomically very flat surfaces besides macroscopic steps. In chemo-mechanical polishing of these silicon wafers bright LEED patterns exhibiting the atomic structure at the actual surface were obtained. No damage or defects could be observed with LEED during this process. This indicates that the removal of Si is mainly controlled by chemical reaction. In an intermediate step the wafers exhibit a slight surface roughness (“haze”) observable by optical Nomarski contrast microscopy or by light scattering. This is correlated qualitatively to a broadening of LEED spots for a phase relation, indicating that the atomic steps are distributed over many atomic layers creating a superimposed long range roughness. After polishing this long range roughness (haze) disappears and atomically smooth surfaces were obtained. Haze free polished Si (111) wafers exhibit on average less than one atomic step for a row of 30 atoms which represents the instrumental resolution. Definitely misoriented wafers show regular terraces. The regularity is increasing with proceeding polishing. The number of atoms per terrace is correlated to the angle of misorientation, demonstrating the high quality of the polished wafer surface.

Rectifying junctions prepared by vacuum deposition of 3,4,9,10-perylenetetracarboxylic dianhydride (PTCDA) and related compounds on both p- and n-type inorganic semiconducting wafers are used for their non-destructive evaluation. By evaporation of metal contact pads onto the organic layer, we can probe many of the fundamental -bulk and surface properties of the semiconductor.

Experimental observation of the photovoltaic effect of a semiconductor grain boundary with a scanning laser beam is analyzed assuming a single trap energy level. The photo-current is calculated from the recombination velocity and the concentration of minority carriers at a grain boundary which was derived from the continuity equation. Open circuit voltage across a sample is obtained from equating this recombination current to compensating thermionic emission current. Using recombination velocity and diffusion length as variables, tha calculated open circuit voltage is compared with experimental data.

The addition of Ta into thin sputtered films of CoCr greatly affects the microstructure and magnetic properties. Grain size, crystallographic preferred orientation and hysteresis parameters have been studied using transmission electron microscopy, electron diffraction, X-ray diffraction, polar Kerr effect and vibrating sample magnetometry techniques. Crystallographic preferred orientation is enhanced and is accompanied by increased perpendicular magnetic anisotropy. An extremely fine-grained microstructure is produced giving rise to films which show no morphological growth features. Thus the observed magnetocrystalline anisotropy is not attributed to columnar growth. Orientation in these films can be further increased by using under-layers of non-ferromagnetic CoCrTa alloys. The mechanism by which this occurs is discussed.

Well characterized rough surfaces of aluminum films have been studied by the nondestructive technique of Spectroscopie Ellipsometry (SE). The roughness of the aluminum specimens had been characterized earlier by Total Integrated Scattering and Stylus Profilometry techniques to obtain numerical estimates of ras roughness and autocovariance lengths. The present SE measurements on these specimens were carried out at a number of angles of incidence in the range 30–80° and at a number of discrete wavelengths in the spectral range 300–650nm. The SE results were then analyzed by the scalar theory of diffraction from random rough surfaces by treating the surface as a simple random rough surface. The results of such analyses of the SE measurements are compared with the results of the earlier characterization techniques.