Abstract
Thin-oxide (40–50 Å) metal oxide semiconductor (MOS) structures are shown to exhibit, prior to large levels of electron tunnel injection, the near-ideal behavior predicted for a uniform trapezoidal barrier with thick-oxide properties. The oscillatory field dependence due to electron-wave interference at the Si/SiO2 interface indicates an abrupt, one-monolayer barrier transition (∼2.5 Å) consistent with earlier work. After tunnel injection of 1017 –5×1018 electrons/cm2, the barrier undergoes significant degradation leading to enhanced tunneling conductance, with reproducible behavior observed among different samples. This effect is consistent with the generation of positive states in the region of the oxide near the Si/SiO2 interface (<20 Å), where the tunneling electrons emerge into the oxide conduction band. Densities of positive-charge and interface-state buildup are also observed from capacitance-voltage (C-V) measurements and are found to be consistent with the observed tunneling dependence on positive-state generation. We suggest that the interface-state buildup may be identical to the positive-state generation observed by tunneling. The generated oxide states are shown to anneal slowly at room temperature, and more rapidly at 100 °C. Comparisons are made between wet, wet/annealed, and dry oxidation processes, with wet oxides exhibiting the largest densities of state generation, and dry oxides the smallest. The results are consistent with other work on the effect of water-related defects in oxides, and with x-ray photoelectron spectroscopy results, showing that breaking of strained Si-O-Si bonds is responsible for state generation during stress. The tunneling data also indicate that, prior to stress, very few positive states reside in the oxide (?1010 cm−2), even though appreciable interface-state desities are observed from C-V data. We therefore conclude that these initial interface states must be associated primarily with the monolayer transition layer and disorder in the underlying silicon, and are not distributed into the oxide near the interface.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have