Large decreases in the conductivity of arsenic-doped silicon have been observed during 500–970°C heat treatments. The rate of conductivity change depends upon the prior quenching rate from diffusion temperature to room temperature. These conductivity changes are reversed by higher-temperature treatments. The relationship between the electrically active arsenic, as calculated from the conductivity, and the total arsenic is shown to be consistent with a model of substitutional arsenic atoms being nonionized when in a cluster or in a complex involving one or more vacancies.

For diffusion of boron into silicon from a doped oxide diffusion source, the surface concentration of boron in silicon and B2O3 in the glass at the glass‐silicon interface are independent of diffusion time. In the temperature range studied, the surface concentrations are also independent of temperature and depend only on the dopant concentration in the deposited glass. This dependence is observed as long as the glass thickness is large enough so that the impurity source approximates a semi‐infinite source. The concentration dependence of the diffusion coefficients of the B2O3 in the glass and the boron in silicon are determined, the diffusion coefficient for boron in silicon being about 100 times that of the diffusion in the oxide. Heating under an oxidizing ambient oxidizes the silicon at the silicon‐glass interface with the thickness growth rate of the undoped SiO2 layer being nearly independent of the original glass thickness. No such interface barrier to the boron diffusion process occurs when argon is used as an ambient.

Some transistor defects show a tendency to occur near the edges of emitters. Like ’’area defects’’, these ’’edge defects’’ occur in silicon wafers in a distribution pattern reminiscent of thermal-stress-induced dislocations. A peculiar feature is that often a number of edge defects which are close replicas of one another occur in a string, each in one of the neighboring emitters. Such edge defects are preponderant in transistors having silicon nitride surface films. In our proposed model, these emitter edge defects are generated when thermal-stress-induced dislocations glide through a row of emitters and interact with the emitter edge stresses, thereby, and under appropriate conditions, casting off small dislocation half-loops which straddle the edges. The model is shown to be consistent with all the observed phenomena pertaining to the emitter edge defect. To provide a foundation for the discussion of the finer details of the emitter edge defects, we also present an analysis of the nature and distribution of thermally induced dislocations, itself a topic of general interest.

We report evidence of the occurrence, and some salient features, of two phenomena arising from ion implantation of silicon: one is a hardening effect of the silicon substrate; the other is a stress-reduction effect in silicon nitride films. The method of indentation dislocation rosettes (IDR) was the tool used for this study. The ion species investigated included boron, arsenic, argon, oxygen, and nitrogen, as well as oxygen from recoil implantation by arsenic into a film of thermal SiO2. Arsenic is the least effective among the ions studied, producing an ∼70% increase in the critical stress-of-dislocation movement (τc), whereas boron brought about an order-of-magnitude increase in τc. Other ions are comparable to boron. A plausible explanation is that the 1×1016 ions/cm2 boron implantation at room temperature was below the critical dose for silicon amorphization (∼2×1016 ions/cm2), so that rapid defect annealing via epitaxial regrowth could not occur, leaving much residual damage. The arsenic implantation of the same dose, however, was well above the critical dose for amorphization (∼2×1014 ions/cm2). In silicon nitride films of various thicknesses, reductions of stresses caused by implanting different ion species were studied with respect to the film-edge-induced stress in the silicon substrates. The results were quantitatively indexed regarding the enhanced movements of IDR near the film edges. Finally, the effects of ion implantation on the yield of bipolar devices are also reported, with the results partly explainable in the framework of the above two phenomena.

No abstract