S. M. Hu

Anomalous temperature effect of oxidation stacking faults in silicon

We observed an anomalous temperature effect in the growth of oxidation stacking faults in silicon. For a given oxidation time, the size of stacking faults first increases with temperature following an Arrhenius relation, reaches a peak at some temperature, and then decreases with temperature rather sharply until, finally, the faults totally vanish. The temperature above which the oxidation stacking faults vanish is dependent on the crystal surface orientation as well as on the oxidation ambients. In dry oxygen, this temperature is ∼1240u2009°C for {100} surfaces, ∼1220u2009°C for {111} surfaces, and ∼1175u2009°C for {1,0,11} surfaces (5° off {100}). Thus, the size‐versus‐temperature curve of the growth of oxidation stacking faults can be divided into two regions, which may be called the growth and the retrogrowth regions. In the growth region the growth follows a power law of (size) ∝ (time)0.8; in the retrogrowth region, the power law breaks down. The activation energy in the growth region is 2.3 eV for all surface ...

Dislocation pinning effect of oxygen atoms in silicon

A definitive proof has now been obtained of the existence of a dislocation pinning effect by oxygen atoms in silicon. Ambiguities caused by uncertain material variables in different crystals or different parts of a crystal are avoided in this investigation by making oxygen out‐diffusion from a wafer, creating a continuous variation of oxygen concentration in a ∼60‐μm‐deep surface layer. Dislocation movement at different depths of this layer was studied using indentation dislocation rosettes (IDR) on a 2° beveled surface. The method of IDR is uniquely suitable for this study because microdislocation half‐loops are generated and confined to move within each ∼10‐μm‐deep layer in a series of precisely determined microregions down the bevel. The size of IDR increases steadily toward the sample surface, in good correlation with the steady decrease of oxygen concentration toward the same sample surface. The effect at the maximum oxygen concentration in this case increases the critical resolved shear stress in si...

Supersaturation of self‐interstitials and undersaturation of vacancies during phosphorus diffusion in silicon

A series of experiments were performed to settle the long‐standing controversy on whether there is a vacancy excess or deficit during the diffusion of phosphorus in silicon. Phosphorus and antimony buried layers sandwiched between p‐type substrates and p‐type epitaxial layers were used to monitor the departure from equilibrium of point defects. It was found that diffusion of the antimony buried layer is retarded during high concentration phosphorus diffusion while diffusion of the phosphorus buried layer is enhanced. This finding is consistent with the hypothesis that there is simultaneously a vacancy undersaturation and a self‐interstitial supersaturation. Possible causes of the interstitial supersaturation are investigated. It is concluded that the process whereby a phosphorus interstitialcy converts to substitutional form by emitting a self‐interstitial is a likely source.

Effects of ambients on oxygen precipitation in silicon

We have discovered an ambient effect on the precipitation of oxygen in the bulk of silicon wafers. Oxidizing ambients were found to strongly retard the precipitation, in comparison with various inert ambients. The precipitation process was studied by the methods of infrared spectrophotometry, thermal conversion, and transmission electron microscopy. In oxidizing ambients, oxygen precipitates tended to cluster in colonies. A typical colony consisted of a few to tens of precipitates, webbed in a complex of dislocation loops. In contrast, precipitates formed in inert ambients were usually isolated. Two mechanisms are suggested to explain this effect, both involving oxidation‐generated silicon self‐interstitials.

Film‐edge‐induced stress in silicon substrates

Stress fields in silicon substrates, induced by edges of silicon nitride films, were studied by experimentally observing the distribution of indentation‐injected dislocation half‐loops in the vicinity of the film edge. From this distribution, the stress field was obtained by applying the method of analysis recently introduced by the author. The results were compared to the theoretical stress field due to a line force (an approximation) tangential to the boundary of a half‐space, with good agreement.

EVIDENCE OF HOLE INJECTION AND TRAPPING IN SILICON NITRIDE FILMS PREPARED BY REACTIVE SPUTTERING

The hysteresis observed in capacitance‐voltage (C‐V) measurements on metal‐sputtered silicon nitride—silicon structures indicates that carriers are injected predominantly as holes rather than electrons. Shifts in the C‐V characteristic after bias‐temperature stress at 300°C support this finding. In dc conduction measurements on these samples a linear relation was found between the logarithm of current and the square root of field. The slope of this plot and the independence of current on bias polarity indicate a bulk‐limited conduction mechanism of emission of carriers from traps in the silicon nitride.

Kinetics of interstitial supersaturation during oxidation of silicon

The kinetics of the supersaturation of self‐interstitials during the thermal oxidation of silicon is reexamined by considering a finite diffusivity of these interstitials. The rate of interstitial generation is assumed to be proportional to the rate of oxidation, and the rate of surface annihilation is assumed to be a first order reaction. The result from an analytical solution shows that within a reasonable oxidation time span, a suitable power‐law kinetics is obtained, with an exponent that falls within the range of most of the reported experimental values.

Point defect generation and enhanced diffusion in silicon due to tantalum silicide overlays

Tantalum silicide deposited directly on monocrystalline silicon substrates and annealed at 950u2009°C causes enhanced diffusion of both boron and antimony in buried layers. The effect is taken as evidence of vacancy supersaturation, since it is known that antimony diffuses in silicon by an almost entirely vacancy mechanism. It also indicates a substantial vacancy component in boron diffusion, at least at 950u2009°C, or lower. The simultaneous occurrence of boron and antimony enhanced diffusion contrasts with the nitridation effect on diffusion previously reported. That the enhanced diffusion occurred in buried layers excludes the snow‐plow mechanism. The Si:Ta ratio of the sputter‐deposited tantalum silicide is slightly less than 2. The interpretation is that further silicidation generates vacancies by removing silicon atoms from the silicon substrate. Enhanced diffusion was not detectable when there was a 150‐nm intervening layer of polycrystalline silicon film between the silicide and the monocrystalline silico...

Growth law for disk precipitates, and oxygen precipitation in silicon

It is shown that the energetically optimum shape of a precipitate which generates strains in the host lattice is a disc with a thickness that increases approximately with the square root of the size (disc radius). This relationship contravenes the two common assumptions that either the thickness or the aspect ratio stays unchanged during precipitate growth. From this relationship is derived a precipitate growth law of r∝t2/3. The results of the analysis are discussed with special reference to the growth of oxygen precipitates in silicon. The change in the morphology of oxygen precipitates with temperatures is also discussed.

New oxide growth law and the thermal oxidation of silicon

A general kinetic law governing the thermal growth of oxide films is presented. It encompasses the Deal–Grove linear‐parabolic growth law and Blanc’s growth law as two special cases; it extends to more general cases where the interface reaction rate has a power‐law dependence on oxidant pressure, such as observed in the thermal oxidation of silicon, where neither the Deal–Grove nor the Blanc growth law is applicable. The new growth law is required even for oxidation under atmospheric pressure because of the varying oxidant pressure at the Si–SiO2 interface. Probable mechanisms for silicon oxidation are suggested.