The IR absorption spectra in the wavenumber region from 500-1200 cm-1 of Si-rich a-Si1-xCx films deposited at various methane gas flows (and therefore exhibiting differ-ent carbon fractions x, see chapter 4.2.3) are depicted in Fig. 5-5 left. The obvious trend for an increasing methane supply in this low power deposition regime is the enhanced incorporation of CH3 groups in the matrix, visible in the increased absorption
Fig. 5-5: Variation of the carbon content (CH4 flow) for as-deposited a-Si1-xCx films depos-ited on c-Si substrate (equivalent behavior on c-Ge substrate). Left: Evolution of the FT-IR absorption spectrum in the wavenumber range from 500-1200 cm-1. Right:
Bond density for integrated Si-H peaks at around 2000 and 2100 cm-1.
64 Passivating a-Si1-xCx films on crystalline silicon and germanium substrates
at around 780 cm-1 related to the Si-CH3 wagging vibration. It is furthermore worth having a closer look at the absorption at around 1000 cm-1: in a pure a-Si1-xCx film (containing Si, C and H atoms only) this feature is assigned to the wagging of CH3
groups. However, although there is a slight increase in absorption with increasing methane gas flow, the existence of this peak cannot be explained in the case of the film with CH4=0. The FT-IR measurements therefore point to oxygen (Si-O-Si stretching with high oscillator strength) present in the a-Si1-xCx matrix. For the occurrence of oxygen, three different positions may be considered, that is, at the crystal-line/amorphous interface (native oxide on the c-Si surface), at the surface of the amor-phous film (oxidation of a-Si1-xCx at air ambient [136]) or oxygen atoms throughout the amorphous bulk. The latter would implicate the presence of oxygen in the reactor during the plasma deposition. However, the low pressures feasible in the setup (<10-6 bar) do not point to leakage of the reactor.
Fig. 5-5 right shows the bond densities for the integrated Si-H peaks at around 2000 and 2100 cm-1 evaluated following the procedure explained in chapter 3.1. The increased CH3 incorporation with increasing methane gas flow is directly correlated with a boost in Si-H bond density which points to the introduction of hydrogenated Si-C entities into the film. This is in accordance with the dissociation process of meth-ane molecules in low power PECVD processes (chapter 4.2.3). More precisely, it is the bond density related to clustered monohydrides (SiH)n and polyhydrides (SiH2)n that undergoes a steady increment whereas the bond density related to isolated monohy-drides SiH decreases. It can therefore be concluded that the films with a higher
Fig. 5-6: Left: refractive index n (633 nm) and extinction coefficient k (400 nm) of a-Si1-xCx
with varying carbon fraction measured by spectral ellipsometry. Right: optical band-gap Eopt extracted from transmission and reflection measurements applying the method proposed by Tauc.
Passivating a-Si1-xCx films on crystalline silicon and germanium substrates 65 C-fraction produced under the given conditions exhibit an increasingly void-rich ma-trix and hence a less dense structure. The latter is confirmed by the optical constants measured for the films with varying carbon fraction, since the refractive index (optical density of the material) as well as the extinction coefficient steadily decrease with increasing methane flow (Fig. 5-6 left). Transmission and reflection measurements of the various films deposited on quartz glass allowed for the evaluation of the optical band-gap Eopt following the definition and method proposed by Tauc [137] (Fig. 5-6 right). An increase in methane gas flow results in a slight but steady widening of the gap. Note that the increase of the optical gap for Si-rich a-Si1-xCx is probably more strongly related to the increased hydrogen incorporation than to the increasing C-content [88]. FT-IR and spectral ellipsometry applied on the respective a-Si1-xCx films deposited on germanium substrate reveal essentially no dependence of the optical and structural properties of the film on the substrate type. Only the deposition rate was found to be higher for (111) Ge than for (100) Si substrates. This is in accordance with the observations of Tweet et al. that the growth rates are essentially determined by hydrogen desorption kinetics [138]. As pointed out in chapter 5.1, hydrogen proceeds faster from a germanium than from a silicon surface therefore probably resulting in an increased growth rate on Ge substrates during the initial stage of deposition.
The evolution of the total Si-H bond density under annealing (30 min at each tem-perature) for various a-Si1-xCx films with differing C-fraction is shown in the Arrhenius representation in Fig. 5-7 left. Independently of the absolute amount of hydrogen
Fig. 5-7: Variation of the carbon content (CH4 flow) in the a-Si1-xCx film. Evolution of total Si-H bond density (left) and effective lifetime (right) of passivated Si samples under anneal-ing in Arrhenius representation. Note that the FT-IR spectra indicate that the a-Si1-xCx
film composition and its evolution under annealing is basically the same for silicon and germanium substrates. Lines are guides to the eye.
66 Passivating a-Si1-xCx films on crystalline silicon and germanium substrates
bonded to silicon atoms, the breaking (reduction) of Si-H bonds starts at temperatures around 300°C and decreases substantially by a factor of 5-10 for annealing tempera-tures of up to 500°C. Tracking of the peak at 780 cm-1for temperatures up to 500°C shows no significant alteration, indicating a constant amount of Si-CH3 groups. The out-diffusing hydrogen therefore predominantly stems from dehydration of silicon atoms. The plotted effective lifetimes τeff for the silicon samples (Fig. 5-7 right) show a direct correlation between the decrease of the total Si-H bond density and the electrical degradation of the passivation quality. In particular, the lifetimes start degrading at a temperature of about 300°C which coincides with the onset of Si-H bond breaking.
The lifetimes related to the sample with CH4=0 (data points omitted for clarity of the graph) are comparable to those with CH4=15 sccm, although the degradation occurring for Tann>300°C is more pronounced as compared to the other samples. Furthermore, the lifetime level decreases with increasing carbon fraction.
While the evolution of the bond densities is practically the same as on silicon sub-strate, the behavior of the measured lifetimes of the germanium samples turns out to be significantly different (Fig. 5-8 left). Most striking is the increased temperature stabil-ity of the a-Si1-xCx passivation on germanium substrates. The measured lifetimes evi-dence an electrical degradation only for annealing temperatures exceeding 450-500°C.
Furthermore, a substantial increase as compared to the as-deposited lifetimes is ob-served for most of the germanium samples under annealing. The dependence on the carbon fraction in the film also differs from the silicon counterpart, exhibiting the
Fig. 5-8: Left: variation of the carbon content (CH4 flow) in the a-Si1-xCx film. Evolution of the effective lifetime of passivated Ge samples under annealing in Arrhenius representation.
Right: onset of electrical degradation of the passivation depending on the carbon con-tent (CH4 flow) in the a-Si1-xCx film for c-Ge and c-Si substrate. Lines are guides to the eye.
Passivating a-Si1-xCx films on crystalline silicon and germanium substrates 67 highest lifetimes (up to 550 µs) in the annealed state for the CH4=45 sccm film. The lifetimes corresponding to the CH4=0 film (data points omitted for clarity of the graph) are comparable to the one with CH4=15 sccm and are significantly lower. The onset temperatures for the degradation of the passivation quality for silicon and germanium substrates are illustrated in Fig. 5-8 right. Note that the onset of Si-H bond breaking is in the range of 300°C for both material types indicating that the passivation mecha-nism of a-Si1-xCx on silicon surfaces is essentially linked to the hydrogen saturation of silicon dangling bonds whereas the mechanism on germanium surfaces is somewhat decoupled from hydrogen.