• Interplay of hydrogen and deposition temperature in optical properties of hot-wire deposited $a\text{‐}\mathrm{Si}:\mathrm{H}$ Films: Ex situ spectroscopic ellipsometry studies

High-quality hydrogenated amorphous silicon $(a\text{‐}\mathrm{Si}:\mathrm{H})$ thin films were grown by hot-wire chemical vapor deposition on glass (Corning 7059) using silane with relatively high hydrogen albeit avoiding the formation of microcrystalline hydrogenated silicon. They were grown as a function of substrate temperature $\left(T_S\right)$ ranging from 50 to 515 °C resulting in the corresponding hydrogen concentration $\left[C_{\mathrm{H}}\right]$ variation from 20.0 to 0.2 at. %. They are optically examined ex situ using spectroscopic phase modulated ellipsometry from near IR to near UV (i.e., 1.5–5.0 eV) obtaining pseudo-dielectric function $(⟨{\epsilon }_r\left(E\right)⟩,⟨{\epsilon }_i\left(E\right)⟩)$ for investigating the role of hydrogen in network disorder. The raw ellipsometry data were modeled using Bruggeman effective medium theory and the dispersion relations for the amorphous semiconductors. A two-layer model consisting of a top surface roughness layer $\left(d_S\right)$ containing an effective medium mix of 50% $a\text{‐}\mathrm{Si}:\mathrm{H}$ and 50% voids and a single “bulk” layer $\left(d_B\right)$ of 100% $a\text{‐}\mathrm{Si}:\mathrm{H}$ was used to simulate the data reasonably well. We performed these simulations by nonlinear least-square regression analysis and it was possible to estimate the true dielectric function, energy band gap $\left(E_g\right)$, film thickness $\left(d_{\mathrm{SE}}\right)$, bulk void fraction, surface roughness layer $\left(d_S\right)$, and confidence limits $\left({\chi }^2\right)$. Moreover, it is shown that the Tauc–Lorentz model fits the ellipsometry data reasonably well and helps elucidating the layered structure of $a\text{‐}\mathrm{Si}:\mathrm{H}$ thin films. We also compared the optical band gap determined using ellipsometry modeling and the Tauc gap. We discuss the variation of the deduced parameters in terms of role of $T_S$ ($T$ role) or of hydrogen ($H$ role) yielding possible physical meaning and found an agreement with the excitation dependent Raman spectroscopy results reported earlier [S. Gupta, R. S. Katiyar, G. Morell, S. Z. Weisz, and J. Balberg, Appl. Phys. Lett. 75, 2803 (1999)]. Atomic force microscopy was also used to validate the simulations. These analyses led to a correlation between the films’ microstructure (or network disorder) and their electronic properties for electronic device applications, in general and for photovoltaic applications, in particular.