No abstract

No abstract

Two models that incorporate the same mobile H+ entity are the radiation model by McLean [I] and the mobile charge model of hydrogen-annealed oxide by Vanheusden et al. [2]. Mobile charge in hydrogen-annealed silicon-on-insulator (SOI) buried oxides before and after irradiation was studied to investigate discrepancies between the two models. We examined Unibond, low-dose SIMOX, and single-and tripleimplant standard-dose SIMOX as well as single-and tripleimplant SIMOX with supplemental oxygen implantation. To measure H+ motion as fast as 0.01 sec we developed a gate pulse method and combined it with standard I-V techniques to measure the full range of H+ response times. All but singleimplant SIMOX exhibit mobile H+ that can be cycled between the Si/Si02 interfaces without reacting or being trapped. Trapping near the top interface attenuates the cycling of H+ in single-implant SIMOX. The transit H+ times were strongly affected by defects in the oxides and varied by an order of magnitude in oxides with the same thickness. The transit time varied linearly with oxide thickness. The effects of irradiation on the mobile H+ was studied to see if the irradiation would introduce defects that modify the H+ behavior and that bring the two models into agreement. No convergence was observed. After irradiation, H+ could be cycled between the Si/Si02 interfaces without reacting and its transit time across the oxide was not altered.

No abstract

Buried oxide (BOX) layers in separated by implantation of oxygen (SIMOX) structures exhibit well‐defined and reproducible bulk (“background”) conduction which, in contrast to localized (“defect”) conduction, is area‐dependent. For annealed samples the bulk conduction is quasi‐ohmic and time dependent up to ∼80 V (for ∼400 nm thick BOX); the temperature dependence indicates 0.3 eV as the activation energy associated with trapping effects. At higher voltage the current‐voltage characteristics are superlinear and resemble those observed for deposited Si‐rich SiO2 and SixNy films, except that they depend on polarity. Fowler‐Nordheim plots indicate an effective barrier height of 1.3 to 1.5 eV for annealed samples and 0.4 eV for unannealed ones; this range is intermediate between 3.1 eV (thermal oxide) and 0.4 to 0.6 eV (Si‐rich SiO2 ). Poole‐Frenkel or Schottky plots false(logI normalvs.E1/2false) are similar to those reported for SixNy films with excess silicon false(x/y >0.75false) . However, the temperature dependence of the high‐field conduction is much less than expected for these mechanisms, indicating that the conduction is controlled by tunneling between Si clusters. From the effective barrier heights of annealed samples the size of the clusters is estimated as ∼0.5 nm (three Si atoms) and their density as 2×1019 cm−3 so that the concentration of excess silicon is at least 6×1019 cm−3 . For unannealed BOX the order of magnitude of excess silicon concentration is estimated as 1021 cm−3. The polarity dependence of conduction is probably associated with the asymmetry in the distribution of excess silicon at the two interfaces. For unannealed samples the current at a given voltage is several orders of magnitude higher than for the annealed ones and the onset of the superlinear regime is at lower voltage; this behavior is similar to the localized (“defect”) conduction exhibited by both annealed and unannealed SIMOX samples. All of the observations indicate that the BOX layers contain excess silicon; this may be in the form of O3normalSi‐SiO3 bonds in the oxide and/or very small silicon (probably amorphous) clusters/filaments.