1 Diffusion in Extrinsic Silicon SFR Workshop & Review April 17, 2002 Hughes Silvestri, Ian Sharp, Hartmut Bracht, and Eugene Haller Berkeley, CA 2002 GOAL: Diffusion measurements on P doped Si to complete comprehensive model of Si diffusion by 9/30/2002.
2 Motivation Reduction of junction depth in CMOS technology To maintain shallow profile: Must control diffusion during further annealing steps (P.A. Packan, MRS Bulletin, June 2000) Understand the Fermi level effect on diffusion and its impact on native defect formation Development of a comprehensive, predictive diffusion model that incorporates physically justified parameters
3 Our Approach Use of an isotopically enriched Si heterostructure allows observation of Si and dopant diffusion simultaneously. Determination of dopant diffusion mechanism and charge state of point defects involved. Determine contribution of charged native defects to Si selfdiffusion. = f I oc I o D I o + f I +C I + D I + + f V C V D V +K ƒ = correlation factor ƒ v = 0.5, ƒ I = 0.73
4 Fermi Level Effect Heavily doped semiconductors - extrinsic at diffusion temperatures Fermi level moves from mid-gap to near conduction (n-type) or valence (p-type) band. Dopant charge state extrinsic n-type + A s extrinsic p-type - A s Fermi level shift lowers the formation enthalpy, H F, of the charged native defect C eq V, I S k F F H kbt ( E E ) o V, I V, I F F = CSi exp exp, H H o V V F V B Increase of C I,V affects Si self-diffusion and dopant diffusion = Native defect charge state I -, V - I +, V + Extrinsic p-type Extrinsic n-type
5 Approach: Isotopically enriched Si heterostructure Alternating layers of natural Si (92.2% 28 Si, 4.7% 29 Si, 3.1% 30 Si) and 28 Si (99.95% 28 Si) (UHV-CVD grown at Lawrence Semiconductor Research Laboratory, Tempe, AZ) Ion implanted amorphous natural Si cap layer used as dopant source to suppress transient enhanced diffusion (TED) (MBE grown at University of Aarhus, Denmark) ( Stable isotope structure allows for simultaneous Si and dopant diffusion studies a-si cap 260 nm 120 nm natural Si 120 nm 28 Si enriched FZ Si substrate
6 Extrinsic p-type: B diffusion experiments Boron dopant source: Boron ion implantation in amorphous Si cap layer: 30 kev, 0.7x10 16 cm -2 37 kev, 1.0x10 16 cm -2 Diffusion: samples sealed in silica ampoules and annealed between 850 o C and 1100 o C 11 B Natural Si layers 30 Si depleted layers 30 Si Secondary Ion Mass Spectrometry (SIMS) used to determine concentration profiles of 11 B, 28 Si, and 30 Si (Accurel Systems, CA) Computer modeling of 11 B and 30 Si concentration profiles and comparison to experimental results
7 Extrinsic p-type: B diffusion results 1000 o C for 4hr 55min 30 Si model fit 30 Si SIMS data B and Si diffusion are described simultaneously Diffusion mechanism (kick-out) B i0 B s- + I 0 + h B i0 B s- + I + Interstitialcy: (BI) 0 B s- + I 0,+ not very likely (f [BI] < 0.3 ) 11 B model fit Si diffusion under intrinsic conditions 11 B SIMS data
8 Extrinsic p-type: B simulation results Supersaturation of I 0 and I + due to B diffusion I 0 and I + mediate Si and B diffusion Enhancement of D B (p) over D B (n i ) under p-type doping Enhancement of D si (p) over D si (n i ) due to Fermi level effect
9 Extrinsic n-type: As diffusion results Implant As into amorphous layer: 0.7 10 16 cm -2 at 130 kev, 1.0 10 16 cm -2 at 160 kev Anneal: 950 C < T < 1100 C Secondary Ion Mass Spectrometry (SIMS) Arsenic sample annealed 950 C for 122 hrs concentration (cm -3 ) 10 21 10 20 10 19 V - V 0 V 2- Reactions for simulation: Vacancy mechanism ( AsV) o As + s + V Interstitialcy mechanism ( AsI) o As + s + I o + e 10 18 10 17 0 200 400 600 800 1000 1200 1400 1600 Depth (nm) V - and I o control Si self-diffusion under n-type extrinsic doping.
10 Extrinsic n-type: As simulation results 10 No supersaturation of I o or V - during As diffusion 10-12 10-13 10-14 D As (ext) C x /C eq 1 V - I o D (cm 2 s -1 ) 10-15 10-16 D As (n i ) 10-17 10-18 (n i ) (ext) 0.1 0 1000 2000 Depth (nm) 10-19 0.7 0.72 0.74 0.76 0.78 0.8 0.82 0.84 0.86 1000/T (1000/K) Enhancement of As and Si self-diffusion under extrinsic conditions entirely due to Fermi level effect.
11 Native defect contributions to Si self-diffusion 10-14 10-15 (n i ) Intrinsic components calculated from extrinsic data D (cm 2 s -1 ) 10-16 10-17 (I o ) (I o +I + +V - ) 10-18 (V - ) (Bracht, et al., 1998) 10-19 (I + ) 0.7 0.8 0.9 1000/T (1000/K) Diffusion coefficients of individual components add up accurately: (n i ) tot = f I oc I o D I o + f I +C I + D I + + f V C V D V = (n i ) (As diffusion) (B diffusion) (As diffusion)
12 Current / Future Developments Isotope heterostructure allows for simultaneous analysis of Si self-diffusion and dopant diffusion. Boron diffusion via the kick-out mechanism: B i0 B s- + I 0,+ As diffusion via interstitialcy and vacancy mechanisms: ( AsI) o As + s + I o + e AsV Able to determine contributions of various native defect charge states to intrinsic Si self-diffusion. Activation enthalpy for self-diffusion due to I o, I +, V - : 2002 and 2003 Goals: ( ) o As s + + V (I 0 ): Q=(4.32 ± 0.15) ev, (I + ): Q=(4.74 ± 0.13) ev, (V - ): Q=(5.18 ± 0.13) ev Diffusion measurements on P doped Si to complete comprehensive model of Si diffusion by 9/30/2002. Apply technique to other materials systems (SiGe), by 9/30/2003.