Diffusion in silicon of elements from columns III and V of the Periodic Table is reviewed in theory and experiment. The emphasis is on the interactions of these substitutional dopants with point defects {vacancies and interstitials) as part of their diffusion mechanisms. The goal of this paper is to unify available experimental observations within the framework of a set of physical models that can be utilized in computer simulations to predict diffusion processes in silicon. The authors assess the present state of experimental data for basic parameters such as point-defect diffusivities and equilibrium concentrations and address a number of questions regarding the mechanisms of dopant diffusion. They offer illustrative examples of ways that diffusion may be modeled in one and two dimensions by solving continuity equations for point defects and dopants. Outstanding questions and inadequacies in existing formulations are identified by comparing computer simulations with experimental results. A summary of the progress made in this field in recent years and of directions future research may take is presented.

Wittmer et al. Reply:In our first paper [l] we reported that converting Pd to Pd2Si induced diffusion of buried marker layers in the underlying silicon substrate. We further reported that the broadening, as determined by secondary ion mass spectroscopy (SIMS) profiling, appeared asymmetric. As we discussed in Ref. [l], while there is no theoretical problem with observing dopant diffusion induced by a surface reaction process, the asymmetric nature of our SIMS profiles is not readily explained by conventional theory.In our second paper [2] we tried to determine if surface topography induced by SIMS sputtering could be solely responsible for profile broadening. To this end, we used two approaches: (l) We made stylus profiling measurements (Alpha Step) in the SIMS craters and imaged the surface topography of the SIMS craters by crosssectional scanning electron microscopy (SEM). The topography information from these two techniques was then compared with corresponding SIMS profiles to ascertain whether or not there was a clear correlation between surface topography and SIMS profile shape. (2) We performed Rutherford backscattering (RBS) measurements on Sb-doped samples in order to obtain information on profile broadening without using the suspect SIMS sputter profiling technique. Our conclusions were as follows: (1) We were unable to discern any clear correlation between the apparent roughness of the surfaces and the deviation of the SIMS profiles from their expected Gaussian shapes. In one case, we observed that the as-grown sample had a rougher surface (after SIMS profiling) than a sample reacted with Pd, even though the sample on which silicidation took place showed a SIMS profile with a large deviation from a Gaussian shape and the as-grown sample had a narrow and symmetric SIMS profile.(2) RBS measurements indicated that Sb buried marker layers were broader after silicidation, though the RBS resolution is not good enough to tell whether the profile is truly asymmetric.Ronsheim and Tejwani [3] repeated our work in an attempt to resolve the question of asymmetric broadening of the measured SIMS profiles. By changing the energy of the analyzing beam to reduce its mixing depth they have been able to remove the asymmetrical broadening of the B marker layer. The fact that the asymmetry of the profiles is affected so strongly by such changes in SIMS conditions is convincing evidence that the asymmetric profiles are an artifact of the SIMS profiling technique. Ronsheim and Tejwani conclude further that the distortion in SIMS profile shapes results from topography induced by the SIMS sputtering process and is not related to a low-temperature diffusion phenomenon. This, of course, is the most logical presumption. However, as we stated in Ref.[2] and reiterate here, in our observations of a large number of samples we could not unambiguously establish a correlation between surface topography and profile shape. This means that the appearance of surface roughness by SEM or standard stylus profiling is not a reliable indicator...

Various phen?~ena ass~ciated with phosphorus diffusion in silicon are reviewed and prominent ~odel~ are cn~lqued. It IS shown that these models are either fundamentally unsound, or are mconslste~t wl!h observed phenomena. A consistent model is proposed in which two mechanisms are operatmg slmult~neous!~, namely, the. vacancy mechanism for the slower diffusing component, and the mterstltlalcy mechamsm for the faster diffusing component. It is assumed that phosphorus exists in silicon in both the substitutional and the interstitialcy species, and that both are shallo~ donors. The c~nversion between the two species is relatively slow, giving rise to the so-called kmked concentratIOn profile. Diffusion via a partial interstitialcy mechanism leads to a supersaturation of self-interstitials.

The changes in diffusion rates of Sb, As, and P resulting from nitridation of SiO2 and direct nitridation of the silicon surface in NH3 ambient at 1100 °C are studied for times ranging from 7 min to 4.5 h. From analysis of these data we conclude that P must diffuse almost entirely by an interstitialcy mechanism at this temperature, and that previous formulations of dopant diffusion under nonequilibrium conditions may not be complete. We also determine that the effects seen during direct nitridation are better explained by a pure vacancy injection process than a pure self-interstitial depletion process, contrary to previous assertions by us and others.

We present the first experimental identification of the diffusion mechanisms of Ge in Si. Using thermal nitridation reactions to create either excess self-interstitials or vacancies, it is established that under equilibrium conditions at 1050 °C Ge diffusion takes place by both substitutional-interstitial interchange and vacancy mechanisms, with comparable contributions from each. If previous conjectures that Ge diffusion in Si is similar to Si self-diffusion are correct, our findings support the idea that Si self-diffusion takes place by both interstitial and vacancy mechanisms.