Abstract
Optical and electron energy-loss data for aluminum between 0.04 and 72 eV have been critically analyzed and used to test the validity of models for (ω), the complex frequency-dependent dielectric constant. Experimental data and models for (ω) can be effectively compared by use of a nonlinear least-squares computer program and, at least in simple cases, the model parameters have physical significance. Though aluminum has been widely regarded as a relatively ideal free-electron metal, it has been found that a Drude model for (ω) does not adequately describe the observed data. Deviations from the Drude model for photon energies greater than about 1 eV have been interpreted in terms of the effects of interband-electronic transitions and a significant L-shell contribution to the real part of (ω) between 10 and 72 eV. From a fit of reflectance data between 0.2 and 12 eV it has been possible to derive parameters describing the interband-transition contribution (b)(ω) to (ω); the imaginary part of (b)(ω) does not differ significantly from calculations based on the aluminum band structure. Optical constants have been derived in the range of fit and agree closely with the measurements of Hass and Waylonis between 1.9 and 5.6 eV. For photon energies less than 0.2 eV, the reflectance data can be fitted by the empirical formulation of Roberts. The optical absorption for photon energies greater than 12 eV is monotonic and greater than that expected from the tails of the electronic transitions at 1.5–2 eV.