ELECTRICAL-PROPERTIES OF HE-IMPLANTATION-PRODUCED NANOCAVITIES IN SILICON

When silicon implanted with > 10(16) He cm-2 is annealed at 700-degrees-C or above, the He forms bubbles and then diffuses out leaving voids (nanocavities). We have prepared n-type and p-type Si samples with nanocavity layers and characterized their structure and their effect on the local band structure and the transport of charge. The ambipolar Si dangling orbitals at the nanocavity walls trap majority carriers in both types of silicon and form quasi-one-dimensional potential barriers which impede transport of charge across the cavity layer. Using dc conductance and high-frequency capacitance techniques we have characterized the height and width of these electrostatic barriers. Capacitance measurements have also been employed to study the evolution of trapped dangling-bond charge as the void containing layers are depleted of carriers in n-type and p-type Schottky-barrier structures. With transient capacitance techniques we have characterized the emission of holes from the unoccupied dangling-bond localized states and also the emission of electrons from the doubly occupied states. The lower dangling-bond level is 0.17 eV above the valence-band maximum, while the upper level lies 0.38 eV below the conduction-band minimum; these energies are qualitatively consistent with broader spectral features observed in ultrahigh-vacuum photoemission experiments on clean reconstructed Si surfaces. Electron paramagnetic resonance (EPR) has been employed to observe unpaired spins associated with dangling orbitals on the cavity walls. The strength of the EPR signal corresponds to approximately 0.1 unpaired spin per dangling orbital, and this reduced amplitude is interpreted in terms of charge redistributions among inequivalent sites which are well known to occur on reconstructed Si surfaces. A simple, one-electron model yields semiquantitative agreement with much of the experimental data on electrical properties and helps explain some of the unusual emission-rate prefactors seen in the capacitance-transient experiments. We nevertheless conclude that a more realistic treatment, including electron-electron repulsion within the cavities as well as charge redistributions on the neutral surface, is probably needed for quantitative prediction.