The basic tenets of the alternative interpretation to be presented here are that the spectral profiles of the star light polarization peaks observed in the visible and near IR are a result of the optical properties of silicate grains in the same spectral range, not of the grain size, provided it remains within the range of Rayleigh’s approximation. The silicate properties are those obtained experimentally by Scott and Duley for the non-iron bearing amorphous forsterite and enstatite. The whole range of observed Serkowski polarization profiles can be simulated with mixtures made of forsterite plus an increasing fraction (0-0.5) of enstatite as the spectral peak shifts from 0.8 to 0.3 µm. Fits to individual observed polarization spectra are also demonstrated. The optical extinction of silicates in the vis/IR (the `transparency range’) can be understood by analogy with the thoroughly studied amorphous hydrogenated carbons and amorphous silica. It is due to structural disorder (dangling bonds and coordination defects) and impurities which give rise to electronic states in the forbidden gap of semiconductors. Because they are partially localized, their extinction power is dramatically reduced and has been ignored or simply described by a low, flat plateau. As their number density depends on the environment, one expects variations in the ratio of optical extinction coefficients in the visible and mid-IR. It is also argued that the measured steep rise of extinction beyond 3 µm-1 into the UV is due to atomic transitions, and so cannot give rise to coherent molecular polarization but only localized extinction.
2018, Monthly Notices of the Royal Astronomical Society, 479, 1685-1693