George H. Bu-Abbud

Variable angle of incidence spectroscopic ellipsometry: Application to GaAs‐AlxGa1−xAs multiple heterostructures

The sensitivity of spectroscopic ellipsometry data to multilayer model parameters is shown to be a strong function of the angle of incidence. A quantitative study of sensitivity versus angle of incidence is performed for a GaAs‐AlxGa1−xAs‐GaAs substrate structure, showing that maximum sensitivity to layer thicknesses and AlGaAs composition occurs near the wavelength‐dependent principal angle. These results are verified by experimental measurements on two molecular‐beam epitaxy grown samples. New spectral features, not found in previous ellipsometry studies of similar structures, are also reported.The sensitivity of spectroscopic ellipsometry data to multilayer model parameters is shown to be a strong function of the angle of incidence. A quantitative study of sensitivity versus angle of incidence is performed for a GaAs‐AlxGa1−xAs‐GaAs substrate structure, showing that maximum sensitivity to layer thicknesses and AlGaAs composition occurs near the wavelength‐dependent principal angle. These results are verified by experimental measurements on two molecular‐beam epitaxy grown samples. New spectral features, not found in previous ellipsometry studies of similar structures, are also reported.

Variable wavelength, variable angle ellipsometry including a sensitivities correlation test

Combining simultaneously both variable wavelength and variable angle of incidence to assess solutions of the optical parameters of a system, a procedure is developed which provides more information on the optical parameters than methods where either the wavelength or the angle of incidence is held constant. This results in a reduction in the interdependence of the optical parameters, and provides for more acceptable solutions. For a film on a substrate structure, if one knows the substrate optical parameters, the thickness and optical properties of the film can be characterized at each wavelength without the need for (a) a dispersion equation, (b) an independent measurement of the thickness or (c) any special requirements on the substrate optical properties. For a nearly transparent film on an absorbing substrate, all four parameters of the three-medium model (the substrate and film refractive indices, substrate extinction coefficient and film thickness) can be determined. If in addition one uses a dispersion relation for the refractive index in a very weakly absorbing film, the optical properties of an absorbing substrate and the film can be determined unambiguously. As a second major feature of this paper we show that the sensitivity correlation matrix, as well as the traditional parameter correlation matrix, should be calculated if one is to determine the independence of ellipsometric parameter solutions. An SiO2 film on silicon is used for illustration, but of course the analysis is recommended for more complicated materials systems.

Parameter correlation and precision in multiple-angle ellipsometry.

A general procedure for the characterization of the optical parameters of a given surface using fixed wavelength multiple-angle-of-incidence ellipsometry is developed. Quantitative evaluation of estimated parameter precision is studied considering the sum of errors squared of a least squares fit, the parameter cross-correlation at this minimum, and the confidence limits of the parameters from variance-covariance relations. Most significantly, it is shown numerically that precision is improved when the solutions exhibit a lower correlation to the remaining parameters when a correct system model is used. Using the silicon dioxide-silicon substrate system we were able to determine the complex index of silicon and the extinction coefficient of the oxide with high precision. Once this was done it was possible to estimate with reasonable confidence the refractive index and particularly the thickness of the oxide despite an observed large cross-correlation coefficient between them. A wide range of starting values for the film refractive index n(f) and thickness t(f) always converged to a small sum of errors squared, and nearly identical values for n(f) and t(f) after accurate estimates were obtained for the other parameters of the system.

Optical and interfacial electronic properties of diamond-like carbon films

Abstract We have been preparing hard semitransparent carbon films on oriented polished crystal wafers of silicon, indium phosphide and gallium arsenide, as well as on KBr and quartz. Properties of the films were determined using IR and visible absorption spectroscopy, ellipsometry, conductance-capacitance spectroscopy and α particle-proton recoil spectroscopy. Preparation techniques include r.f. plasma decomposition of methane (and other hydrocarbons), ion beam sputtering and dual-ion-beam sputter deposition. Optical energy band gaps as large as 2.7 eV and extinction coefficients lower than 0.1 at long wavelengths are found. Electronic state densities at the interface with silicon as low as 10 10 states eV -1 cm -2 were found.

Mass density and hydrogen concentration in diamond-like carbon films: proton recoil, rutherford backscattering and ellipsometric analysis

Abstract We used helium-induced hydrogen recoil analysis to determine the concentration of hydrogen versus the anneal temperature in r.f. plasma-deposited “diamond-like” carbon films. In addition, Rutherford backscattering combined with spectroscopic ellipsometry was used to determine simply, accurately and non-destructively the mass density of the films. By comparing with “standard” samples, the hydrogen concentration is found to be as high as 47 at.% depending on the deposition conditions. The hydrogen concentration drops rapidly with anneal temperatures up to 960°C. The densities of three films prepared by r.f. plasma deposition were (1.7–1.8) × 10 3 kg m −3 . This value is in good agreement with values found by conventional techniques. It is likely that hydrogen plays a major role in determining microstructure, bonding and density.

An enhanced sensitivity null ellipsometry technique for studying films on substrates - Application to silicon nitride on gallium arsenide

General considerations are applied to optimize the sensitivity of ellipsometric measurements for thin films on a substrate. s‐ or p‐wave suppression conditions are found to give maximum sensitivity. Approximate values of the optical parameters of the films and substrate are used to calculate discrete film thicknesses for the s‐ or p‐wave suppression to occur. For null fixed‐wavelength ellipsometry, these calculations are limited to experimentally available wavelengths, e.g., at strong emission lines from a Hg lamp. Films with thicknesses near the calculated ones are then deposited on the substrate. The ellipsometric parameters ψ and Δ are obtained at multiple angles of incidence and wavelengths, and a least‐squares procedure is used for the analysis. The method has been applied to silicon nitride films on GaAs. The problem of correlation between the calculated optical parameters of the system is addressed. It is shown that the multiple‐wavelength analysis decreases significantly the correlations as compar...

Roughness studies of ion beam processed molybdenum surfaces

Mechanically polished molybdenum samples have been irradiated with 150 keV molybdenum ions at fluences from zero to 16×1015 ions/cm2 to study the effects on surface smoothing. Both fluence and substrate temperature during irradiation (25, 250, and 500 °C) had considerable effects on the optical properties. Ellipsometry, profilometry, and Nomarski photography were used to characterize the surfaces. Multiple wavelength, multiple angle of incidence ellipsometry results were analyzed using effective medium approximation models, with molybdenum, molybdenum oxide, and voids as the principal constituents. Generally, we find that the reflectance calculated from the ellipsometric measurements increases continuously as a function of fluence. However, for high substrate temperature and short wavelength the reflectance as a function of fluence reaches a maximum and then decreases. Effective medium calculations indicate that there is a corresponding change in the surface roughness with fluence, and that effects due to...

Analytic solutions for optimized ellipsometric measurements of interfaces and surface layers in thin film structures

Abstract The maximum experimental sensitivity in ellipsometric measurements for samples that include a composite (two-layer) film on a substrate is treated with mathematical generality. A detailed theoretical computation is made for a system consisting of a moderately thick film on a substrate and including either an interface layer between the film and substrate or a thin surface layer on top of the moderately thick film. Analytical conditions for the sensitivity of p or s wave suppression are determined. For the air/SiO 2 /Si system a numerical calculation is performed to determine the optimum sensitivity for interfacial layer studies.

Study of Mo‐, Au‐, and Ni‐implanted molybdenum laser mirrors by multiple angle of incidence spectroscopic ellipsometry

Multiple angle of incidence spectroscopic ellipsometric data show that implantation of 150‐keV molybdenum ions into polished molybdenum laser mirrors causes microscopic surface smoothing, and that most of the microscopic roughness is removed by a fluence of 5×1015 cm−2. Implantation of Au at 1 MeV significantly increases the microscopic roughness, and also changes the bulk optical properties. 3‐MeV Ni ion implantation causes only small changes in the surface and bulk properties. A dielectric film, probably a hydrocarbon, is found to condense on the mirrors in a laboratory atmosphere.

Optimizing null ellipsometry for oxidized silicon

Polarizer imperfections which produce a deviation of the measuring light beam cause systematic errors in the estimation of optical parameters particularly for low absorbing systems such as oxidized silicon. These can be reduced considerably by using the polarizer, specimen, compensator, analyzer (PSCA) instead of the PCSA arrangement. In addition, PSCA measurements can be made nearer to the principal angle where sensitivity is greatest. The discrepancy of Delta measurements between zones is reduced by a factor of 20. Precision is improved by a factor of 10-50 for the PSCA measurements near the principal angle compared with measurements at any angle of incidence for the PCSA arrangement. Results on silicon covered with a natural oxide are compared for illustration.