Quantitative theory of retarded base diffusion in silicon NPN structures with arsenic emitters

When arsenic (As) diffused into gallium (Ga) or boron (B) doped Si, a retardation of the p-type base layer is generally observed; this is in contrast to the "emitter-push" effect associated with phosphorus diffusions. In order to simulate transistor profiles, it is necessary to be able to quantitatively describe the emitter-base interactions during diffusion. In this study, the way in which the internal electric field, the equilibrium vacancy density, ion pairing and the rate of (VsiAS2) complex formation affect the redistribution of the base layer during sequential processing was investigated. Numerical solutions to the coupled diffusion equations indicate that the electric field and ion pairing effects only cause localized retardation of a B profile during the As emitter diffusion. However, the formation of (VsiAs2) complexes causes a vacancy undersaturation in the Si to a distance in the crystal well beyond most practical collector-base junction depths. Since the local base diffusivity depends upon the vacancy density, this extrinsic vacancy undersaturation effect causes the expected retarded base diffusion. Experimental verification of the theory presented is given. It is also shown that the retardation, δ, has the following functional dependences on the parameters listed below: 1. Emitter diffusion temperature, decreases with increasing TE(Above 1200°C, no measurable retardation exists for practical transistor junction depths). 2. Emitter diffusion time, increases with increasing tE. 3. Arsenic surface concentration, C2(O,t)---δ increases with increasing C2(O,t) (No measurable retardation for C2(O,t) cm^-3for practical diffusion times). 4. Initial base depth, XCBO---δ increases with decreasing XCBO(No measurable retardation for depths greater than one vacancy diffusion length). 5. Initial base surface doping, increases with increasing C1(O,O) (No measurable retardation for C1(O,O) cm^-3).