A Simplified Picture for Transient Enhanced Diffusion of Boron in Silicon

Abstract

In recent years, transistor junction formation in complementary metal oxide semiconductor devices by ion implantation has encountered serious limitations due to transient enhanced diffusion ͑TED͒ during the annealing step. Current models of TED rely heavily on detailed simulations of the complex diffusion-reaction network that governs TED, and often rely on fitted parameters whose values are uncertain. The present work uses a more rigorous set of rate parameters obtained from a maximum likelihood estimation to develop a relatively simple analytical treatment of boron TED that is capable of estimating the degree of profile spreading and the temperature at which TED should begin to occur significantly. The treatment suggests that reduction of TED should focus on implantation schemes and heating programs designed to decrease the number of clusters slightly smaller than the very largest.Forming extremely shallow pn junctions in Si-based microelectronic logic devices is becoming increasingly critical as device dimensions continue to diminish. For example, advanced complementary metal oxide semiconductor devices will require junction depths X j between 13 and 22 nm in the source and drain extension regions by 2005. 1 Current technology for junction formation relies almost exclusively on ion implantation to introduce dopants into the substrate. Although junction depths can be made shallower by reducing the implant energy, the effectiveness of this approach has been limited by the need to anneal the resulting structure to activate the dopant electrically and to eliminate implant-induced defects in the crystal structure. As long as they exist, these defects mediate exceptionally fast transient enhanced diffusion ͑TED͒ of the implanted dopants, often leading to significant spreading of the original dopant profile.Because experiments to measure TED kinetics are expensive and sometimes difficult to interpret unambiguously, many researchers have resorted to detailed modeling to aid process development. There have been several attempts in the literature to develop a comprehensive physical picture for TED, 2-5 and such attempts have been incorporated into various widely used profile simulators. 6 However, the existing models suffer important deficiencies.One problem is that the predictive capability of most TED models outside their tested range is subject to serious doubt. Many elementary kinetic steps contribute to the experimental observable: typically a dopant depth profile obtained by secondary ion mass spectroscopy ͑SIMS͒. Hence, all models include numerous rate parameters, many of whose values are developed primarily according to their ability to fit experimental SIMS profiles. The large number of parameters provides many degrees of freedom for fitting, impeding the ability to develop a unique set.A second problem is that profile simulators are not very conducive to developing simple, intuitive explanations for key features of TED that retain quantitative utility. Many kinetic processes take place in ways that vary strong...

Structure and stability of small compact self-interstitial clusters in crystalline silicon

“…It is now well accepted that the diffusion and clustering of implanted dopants are mainly mediated by native defects created during dopant introduction. In particular, excess Si interstitials are mainly responsible for boron TED and deactivation [1][2][3][4][5][6][7][8][9][10][11][12][13] and also play a critical role in TED and clustering of arsenic 1,5,14-20 and phosphorous. 5,19 Single interstitials are highly mobile even at room temperature.…”

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

“…38,39 These mobile species can take on multiple charge states in response to changes in dark Fermi level 40 as well as photostimulation. 15 Thus, in principle it is possible that photostimulation alters these charge states, which could propagate into changes in diffusivity that are manifested in the present results.…”

First-Principles Prediction of Optical Absorption Enhancement for Si Native Defect Clusters under Biaxial Strain

“…Improved descriptions [32][33][34] of the aggregation and behavior of Si point defects following ion implantation were fueled by the development of microelectronic fabrication process simulators. Previous theoretical works by Makhov and Lewis 35 ͑V 4 ͒ and Arai et al 36 ͑I 4 ͒ proposed the importance of FC configurations for specific cluster sizes ͑n͒.…”