Time-resolved optical studies of oxide-encapsulated silicon during pulsed laser melting
- PDF / 1,016,671 Bytes
- 8 Pages / 593 x 841.68 pts Page_size
- 2 Downloads / 179 Views
Nanosecond time-resolved reflectivity and ellipsometry experiments have been performed on (100) Si wafers encapsulated by 5.5-76.2 nm thick thermal oxides, using pulsed KrF (248 nm) laser energy densities sufficient to melt the Si beneath the oxide. Post-irradiation nulling ellipsometry, optical microphotography, and surface profiling measurements were carried out. It was found that the threshold energy density required to melt the Si varies with oxide thickness; this is explained primarily by the reflective properties of the oxide overlayer. The time-resolved reflectivity and ellipsometry measurements show that rippling of the SiO2 layer occurs on the 20-40 ns timescale and results in a decrease in specular reflectivity of the rippled silicon surface beneath. Optical model calculations suggest that pulsed laser annealing through a thick oxide layer results in a damaged near-surface silicon layer ( ~ 30 nm thick); this layer contains defects that are probably responsible for the degraded performance of devices. I. INTRODUCTION Pulsed laser annealing (PLA) of semiconductors has been the subject of many investigations over the last eight to ten years.' The basic process is now well understood: If a crystalline semiconductor is irradiated with pulsed light of energy density E,, then the sample surface will melt when El exceeds a certain threshold value. A melt front will propagate rapidly into the sample, stop, and a solidifying front will propagate back to the sample surface at a slower velocity. If the near-surface region of a crystalline semiconductor has been amorphized by ion implantation and the laser E, is sufficient to melt entirely through the amorphous layer, then high velocity epitaxial regrowth occurs. In the early days of PLA studies, it was thought that PLA might become very useful for the fabrication of complex silicon integrated circuit (IC) structures, since the PLA process heats only the top few fim of the sample, and formation of very shallow and abrupt doped regions is possible. However, many of the proposed uses of PLA in IC fabrication involved irradiation of a silicon wafer through a thick thermal oxide. Initial experiments on PLA of thermally oxidized silicon2"6 revealed significant problems: (1) The oxide tended to ripple, (2) voids and dislocation loops were found near the silicon-silicon dioxide interface, and (3) the oxide layer tended to move laterally near any welldefined oxide edge. Further experiments showed that the oxide ripple spacing was dependent upon its thickness and, to a lesser extent, upon the Et and temporal shape of the laser pulse. The threshold E, for rippling was very close to the melting threshold. Rippling was observed only on the oxide-covered parts of the silicon wafer, and the ripple pattern was found to be embedded in the silicon itself, since it remained after the oxide was 498
J. Mater. Res. 3 (3), May/Jun 1988
stripped with HF. Furthermore, TEM studies revealed increased etch pit and dislocation densities near the Si-SiO2 interface; these resulted in poor devic
Data Loading...
