Surface control of interstitial behavior for improved ultrashallow junction formation

Abstract

There is increasing evidence that surface proximity effects must be incorporated into models for transient enhanced diffusion (TED). The present work examines the previously unrecognized influence that near-surface band bending can have on dopant profiles. Experiments employ the optical technique of photoreflectance to show that band bending exists at the Si-Si02 interface just after implantation. The effects of such band bending are investigated numerically using a simulator whose rate parameters have been developed from literature data using Maximum Likelihood (ML) estimation together with multivariate statistics to quantify accuracy. The resulting simulator yields excellent tits of SIMS profiles with no freely adjustable parameters, and shows that band bending transforms interfaces into reflectors of charged interstitials (i.e., no flux), even if the interface would otherwise serve as a good sink for these defects. This transformation deepens the junction significantly and also induces the pileup of dopant very close to the interface.

Interface-Mediated Photostimulation Effects on Diffusion and Activation of Boron Implanted into Silicon

“…Some progress was achieved in quantifying photoenhanced diffusion in II-VI semiconductors (using computational methods) 9 and a-Si:H (experimentally), 10 yet the understanding of the photostimulated dopant diffusion in c-Si remained inadequate. This laboratory has employed a novel experimental configuration to eliminate the ambiguity, and has thereby demonstrated the influences of low-intensity (<1 W/cm 2 ) super-bandgap illumination − first on the surface diffusion of various adsorbates on Si, 11,12 and subsequently on the bulk self-diffusion of Si 13,15 as well as the bulk diffusion and electrical activation of As and B implanted into silicon. 14 As first suggested for surface diffusion 11,12,15,16 and later modeled quantitatively for bulk Si self-diffusion, 13 photostimulation can influence diffusional flux rates by changing the average electrical charge state of point defects involved with diffusion, although the exact mechanism depends upon the specific case.…”