10.8(d3) shows a spheroidization effect for the smaller diameter nanowires, which
is attributed to annealing of the nanowires, as will be described later.
In region II, the OAG region, the diameters of the Si nanowires are quite uniform, irrespective of the substrate temperature. This may be explained by a vapor–
solid (VS) process governing the initial nucleation of the OAG rather than a VLS
process in which the size of the liquid droplet decreases with decreasing energy,
leading to a respective decrease in the SiNW diameter as discussed above (for the
metal catalyst growth). This would mean that the small nuclei of crystalline Si
nanowires were directly solidified from SiO in the vapor phase. This explanation is,
however, not in accord with the formation of a SiO x cap on the top of the SiNW
and the alternative proposition that the oxide-assisted nucleation and growth is occurring due to: (1) the lower melting point of SiO x compared to that of SiO 2 , (2)
the high reactivity of the molten SiO x cap to Si-containing clusters in the vapor,
and (3) the decomposition of Si suboxide to Si and SiO 2 . We proposed that the
Si n O m clusters react with the SiO x cap, the crystalline Si core precipitates below
and the excess oxygen diffuses to the sides forming a SiO 2 amorphous outer layer
which solidifies, due to its higher melting point, and limits the further lateral
growth of the nucleus. The lateral growth results from the energetically favorable
adsorption of vapor clusters by the highly curved SiO x molten tip on the one hand
and the lateral restriction imposed by the solid SiO 2 sheath on the other hand (as
shown in Figure 10.9(d2)). The SiNW diameter may be determined not only by the
diameter of the initial SiO x droplet, but also by the equilibrium between the condensation and the disproportionation of SiO to Si and SiO 2 and by diffusion of the
excess O to the sides. It is still not completely understood why this equilibrium is
not temperature-dependent under our experimental conditions. It could be that the
dependence is weak in this limited temperature region and will be revealed if we
enlarge this region by using different experimental conditions.
The SiO vapor phase is stable at a high temperature, so that the condensation
and disproportionation of the SiO vapor into Si þ SiO 2 occurs only below a certain
substrate temperature (the upper limit of region II). On the other hand, below the
lower limit in region II the SiO vapor condenses directly to form SiO solid [59],
with no preferential adsorption nor disproportionation, so that the 1D growth is
suppressed. This explains why the OAG of SiNWs was restricted to the temperature range 1100–850
C (region II).
Finally, we discuss the formation of tadpole-like and chain-like Si nanostructures
from the metal-catalyst VLS and the OAG (Figures 10.8 and 10.9) processes. Both
can be described in terms of a spheroidization mechanism. One-dimensional
SiNWs are less stable than the three-dimensional bulk Si, since the wire has a
much larger surface area and thus higher surface energy. Annealing of SiNWs for
a sufficient time results in spheroidization, as shown schematically in Figure
10.10. The chemical potential of the SiNWs varies with the local curvature so that
small variations in their diameter generate a driving force for diffusive transport
between different chemical potentials. The Si nanowire would convert into a
nanospherical chain first (as shown schematically in Figure 10.10 and experimentally in Figure 10.9(b)). Later, with further diffusion, the inner crystalline Si core
10 Oxide-Assisted Growth of Silicon and Related Nanowires
322
is attributed to annealing of the nanowires, as will be described later.
In region II, the OAG region, the diameters of the Si nanowires are quite uniform, irrespective of the substrate temperature. This may be explained by a vapor–
solid (VS) process governing the initial nucleation of the OAG rather than a VLS
process in which the size of the liquid droplet decreases with decreasing energy,
leading to a respective decrease in the SiNW diameter as discussed above (for the
metal catalyst growth). This would mean that the small nuclei of crystalline Si
nanowires were directly solidified from SiO in the vapor phase. This explanation is,
however, not in accord with the formation of a SiO x cap on the top of the SiNW
and the alternative proposition that the oxide-assisted nucleation and growth is occurring due to: (1) the lower melting point of SiO x compared to that of SiO 2 , (2)
the high reactivity of the molten SiO x cap to Si-containing clusters in the vapor,
and (3) the decomposition of Si suboxide to Si and SiO 2 . We proposed that the
Si n O m clusters react with the SiO x cap, the crystalline Si core precipitates below
and the excess oxygen diffuses to the sides forming a SiO 2 amorphous outer layer
which solidifies, due to its higher melting point, and limits the further lateral
growth of the nucleus. The lateral growth results from the energetically favorable
adsorption of vapor clusters by the highly curved SiO x molten tip on the one hand
and the lateral restriction imposed by the solid SiO 2 sheath on the other hand (as
shown in Figure 10.9(d2)). The SiNW diameter may be determined not only by the
diameter of the initial SiO x droplet, but also by the equilibrium between the condensation and the disproportionation of SiO to Si and SiO 2 and by diffusion of the
excess O to the sides. It is still not completely understood why this equilibrium is
not temperature-dependent under our experimental conditions. It could be that the
dependence is weak in this limited temperature region and will be revealed if we
enlarge this region by using different experimental conditions.
The SiO vapor phase is stable at a high temperature, so that the condensation
and disproportionation of the SiO vapor into Si þ SiO 2 occurs only below a certain
substrate temperature (the upper limit of region II). On the other hand, below the
lower limit in region II the SiO vapor condenses directly to form SiO solid [59],
with no preferential adsorption nor disproportionation, so that the 1D growth is
suppressed. This explains why the OAG of SiNWs was restricted to the temperature range 1100–850
C (region II).
Finally, we discuss the formation of tadpole-like and chain-like Si nanostructures
from the metal-catalyst VLS and the OAG (Figures 10.8 and 10.9) processes. Both
can be described in terms of a spheroidization mechanism. One-dimensional
SiNWs are less stable than the three-dimensional bulk Si, since the wire has a
much larger surface area and thus higher surface energy. Annealing of SiNWs for
a sufficient time results in spheroidization, as shown schematically in Figure
10.10. The chemical potential of the SiNWs varies with the local curvature so that
small variations in their diameter generate a driving force for diffusive transport
between different chemical potentials. The Si nanowire would convert into a
nanospherical chain first (as shown schematically in Figure 10.10 and experimentally in Figure 10.9(b)). Later, with further diffusion, the inner crystalline Si core
10 Oxide-Assisted Growth of Silicon and Related Nanowires
322