The microstructural features of the Si-SiO? system and the chemical physics of its defects are reviewed and examined. Topics are grouped by scientific commonality, rather than by the usual technological manifestations. The role of atomic and molecular sized entities is emphasized, and the latter are limited to those containing only Si, 0, H, or combinations thereof. Most of the reported researches involve x-ray or electron diffraction, Auger or photoelectron spectroscopy, Rutherford backscattering, electron spin resonance, or capacitance-voltage or deep-level transient spectroscopy. Several forms of crystalline and amorphous vitreous silica are considered as a basis for discussion of thin film thermal silica on silicon wafers. Local lattice symmetry, stoichiometry, bond lengths and angles, vacancies and voids, dangling orbital centres, and fixed and migratory hydrogen species are treated extensively. Elements of relevant theory are summarized. Overall, it is hoped to provide a solid data base for future development of general models for essential electronic phenomena in the Si-Si02 system.

Interface states and electron spin resonance centers have been observed and compared in thermally oxidized (111) and (100) silicon wafers subjected to various processing treatments. The ESR Pb signal, previously assigned to interface ⋅Si≡Si3 defects on (111) wafers, was found to have two components on (100): an ⋅Si≡Si3 center oriented in accord with (100) face structure, and an unidentified center consistent with ⋅Si≡Si2O. The quantitative proportionality of Pb spin concentration to midgap interface trap density Dit is maintained on (100), and both are lower by a factor of about 3 compared to (111). This correlation persists over the range of oxidation temperatures 800–1200°C, for both n- and p-doped silicon, cooled by fast pull in oxygen, and cooled or annealed in nitrogen or argon. The correlation is independent of doping level. In samples with different oxide thickness, neither Pb nor Dit varied significantly over the range 100–2000 A, but Pb was smaller at 50 A. In general, ESR is judged to offer promise for further studies of specific interface features.

The band-gap energy distribution of Pb centers on oxidized (100) Si wafers has been determined and compared with interface electrical trap density Dit. Two different Pb centers are observed on (100) Si: Pb0, which has the structure ⋅Si≡Si3, and is essentially identical to the sole Pb center observed on (111) Si; and Pb1, of presently uncertain identity, but clearly different in nature from Pb0. By electric field-controlled electron paramagnetic resonance (EPR) and capacitance-voltage (C-V) measurements, it is found that Pb0 has its (0↔1) electron transition at Ev+0.3 eV and its (1↔2) transition at Ev+0.85 eV. Similarly, Pb1 has its (0↔1) transition at Ev+0.45 eV and its (1↔2) transition at Ev+0.8 eV. The Pb band-gap density correlates qualitatively and quantitatively with the electrical trap density Dit from C-V analysis; nonbonded Pb orbitals are found to be the source of about 50% of the characteristic traps in dry-oxidized, unannealed (100) Si wafers.

Energy distribution of Pb centers (⋅Si≡Si3) and electronic traps (Dit) at the Si/SiO2 interface in metal-oxide-silicon (MOS) structures was examined by electric-field-controlled electron paramagnetic resonance (EPR) and capacitance-voltage (C-V) analysis on the same samples. Chips of (111)-oriented silicon were dry-oxidized for maximum Pb and trap density, and metallized with a large MOS capacitor for EPR and adjacent small dots for C-V measurements. Analysis of C-V data shows two Dit peaks of amplitude 2×1013 eV−1 cm−2 at Ev+0.26 eV and Ev+0.84 eV. The EPR spin density reflects addition or subtraction of an electron from the singly occupied paramagnetic state and shows transitions of amplitude 1.5×1013 eV−1 cm−2 at Ev+0.31 eV and Ev+0.80 eV. This correlation of electrical and EPR responses and their identical chemical and physical behavior are strong evidence that ⋅Si≡Si3 is a major source of interface electronic traps in the 0.15–0.95 eV region of the Si band gap in unpassivated material.

The ESR Pb center has been observed in thermally oxidized single-crystal silicon wafers, and compared with oxide fixed charge Qss and oxidation-induced interface states Nst. The Pb center is found to be located near the interface on (111) wafers. Its g anisotropy is very similar to that of known bulk silicon defects having SiIII bonded to three other Si atoms; the Pb unpaired electron orbital, however, is exclusively oriented normal to the (111) surface. The Pb center cannot be identified with any other known defect in Si or SiO2; in particular, it is totally unlike the common E′ center of SiO2. In contrast to Qss, both Pb and Nst were found to be greatly reduced by steam oxidation and hydrogen annealing. Both Pb and Nst may be regenerated by subsequent N2 anneals at 500 °C. In a graded series of samples, Pb and Nst are found to be proportional and nearly equal in concentration. This possible confirmation of SiIII at the interface, and correlation with Nst, support the theoretical indication of an SiIII band-gap energy level. The E′ center is unobservable, and if present, exists only in a concentration well below that of Qss. Thus, in addition to a lack of strong correlation with Pb, Qss is evidently not due to E′ centers in their normal charge state. Overall, ESR is judged to be a useful technique for research on silicon wafer defects.

Although negative-bias-temperature instability in metal-oxide-semiconductor integrated circuits has been minimized empirically, the exact mechanism is unknown. We argue in this paper that the mechanism of negative-bias-temperature instability can be modeled by a first-order electrochemical reaction between hydrogenated trivalent silicon, a neutral water-related species located in the oxide near the Si-SiO2 interface, and holes at the silicon surface to form neutral trivalent silicon and a positively charged water-related species. To show that such a reaction describes the phenomenon, we show that (1) water must be present in the oxide near the Si-SiO2 interface, (2) induced interface and oxide-fixed charge densities are equal, (3) the saturation interface-trap and oxide-fixed charge densities depend on the initial hole concentration at the silicon surface or aging field, (4) the buildup of these charge densities follows first-order reaction kinetics, and (5) time constants for this charge buildup are independent of aging field. The measurements which are done to demonstrate these features combine room-temperature charge measurement using the Q-C method with current measurements during accelerated aging.

The metal-oxide-semiconductor (MOS) transistor has become the dominant device for very large-scale integrated circuits. The performance and reliability of an MOS device are heavily influenced by the quality and properties of the interface between the oxide and the Si region directly beneath. Inherent, process-related, and operationally and environmentally generated interface states or traps are exceedingly harmful or disabling when present. Although controlled successfully by semiempirical design, fabrication and operational regimens, ever-smaller device size-approaching the scale in which a single trap can be important-makes further extension of knowledge on the basic physical and chemical aspects of interface states essential for confident technological progress. In addition, Si/SiO, interface states continue to merit research in their own right. This paper briefly highlights selected topics from past and present interface-state research, describes some essential physical and chemical properties, and introduces subjects to be covered in ensuing specialised papers.

We have explored the nature of the silicon dangling-bond center in amorphous hydrogenated silicon nitride (a-SiNx:H) thin films, and its relationship to the charge trapping centers using electron paramagnetic resonance (EPR) and capacitance-voltage (C-V) measurements. We have investigated the quantitative relationship between the concentration of silicon dangling bonds using EPR and the concentration of charge traps, measured by C-V measurements, for both UV-illuminated and unilluminated a-SiNx:H thin films subjected to both electron and hole injection sequences. A theoretical framework for our results is also discussed. These results continue to support a model in which the Si dangling bond is a negative-U defect in silicon nitride, and that a change in charge state of preexisting positively and negatively charged Si sites is responsible for the trapping phenomena observed in these thin film dielectrics.