A new model for Boron diffusion retardation in SiGe-strained layers accounting for the mechanism of Boron trapping/detra
- PDF / 99,401 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 12 Downloads / 165 Views
A new model for Boron diffusion retardation in SiGe-strained layers accounting for the mechanism of Boron trapping/detrapping by Ge atoms. Victor I. Kol’dyaev PDF/Solutions, Inc (333 San Carlos, San Jose, 95110, CA, USA) ABSTRACT The main drawbacks of the known models of the B diffusion in strained SiGe layers are summarized. A mechanism is suggested to self-consistently explain the main experimental features and original experimental data which considers the trapping of B atoms by Ge atoms during B diffusion in the Si lattice resulting in the retarded B diffusivity. Fluctuations of Ge atom numbers in a nearest B atom environment result in percolation mechanism of B transport through dilatation centers of random size. A new solid state transport model is generalized by considering dispersion transport of positive and negative point dilatation defects. INTRODUCTION Physics based TCAD models are vital in developing a new HBT technology and device architecture. The ready-made TCAD tools for the technology simulation are not accurate for the modern SiGe HBT (“SiGe” is used standing for Si1-xGex where x is the atomic percentage of Ge in a layer). In contrast to known B diffusion physics in Si [1] new principal physical phenomena of a strong B diffusion retardation in strained SiGe, less retardation in relaxed SiGe and B segregation into SiGe are observed [2-15] in SiGe. B in-diffusion into SiGe layer is suggested as a new method for an experimental estimation of B segregation into SiGe. Experiments based on the method are carried out to estimate the segregation coefficient which is higher than observed in [9-11]. There are 2 different models for B diffusion in SiGe [6,9-11]. Model [6] deals with a strain effect coefficient Q as = dE A / ds being characterized for Si(Ge) and SiGe(B) systems. One has: D A = D A0 exp(− E A / kT ) * exp(+ Qas s / kT ) , where s is the coherent local strain due to the Ge presence. In [6] QGes =40±5 eV/unit strain (u.s.), E A =5.3325 eV and D A0 =1.03 105 cm2/sec are found for the low content Ge diffusion in Si. Also Q Bs =-17±3 eV/u.s. is extracted for B in SiGe [6]. This is a simple, accurate and useful model describing a lot of features of B and Ge diffusion in SiGe. The model predicts a monotonic dependence of B diffusivity vs. strain and monotonic B profiles at the Si-SiGe interface resulting from the condition: J B = − DB ( Si )∇C B ( Si ) = = − DB ( SiGe )∇C B ( SiGe ) = const ( x b ) which is not observed in the experiments. The model does not explain the “segregation” phenomena [9-11] and diffusion against the B gradient [4-5,12]. In model [9-11] the segregation phenomena was attributed to the B electrochemical potential difference between Si and SiGe depending on the DOS, the work function, the binding energy, and the lattice contraction energy difference between Si and SiGe. The lattice contraction energy E ca of B in SiGe is taken: E ca = θβ B β Ge C Ge , where θ is a constant, β B and β Ge are the B and Ge lattice contraction coefficients in Si, correspondingly, and C Ge is a Ge c
Data Loading...
