Barrier heights at the polycrystalline silicon-SiO₂ interface

Abstract

Capacitance-voltage measurements of the dependence of flat-band voltage of polycrystalline-silicon (polysilicon) gate metal-oxide-silicon capacitors on oxide thickness are used to measure the position of the Fermi level in degenerately doped polysilicon as a function of the doping of the polysilicon. For boron-doped polysilicon, the experimental difference between the Fermi level in polysilicon and the intrinsic level in the substrate silicon, φms, is ∼0.54 V. The Fermi level is pinned within 20 meV of the valence band edge. For arsenic-doped and phosphorus-doped polysilicon, evaluation of φms is complicated by the occurrence of positive charge at the polysilicon-SiO2 interface which is generated by reaction of doped polysilicon with SiO2. If such positive charge is close to the polysilicon-SiO2 interface it acts as a dipole layer and lowers the values of φms. If positive charge is slightly deeper in the SiO2 it contributes to negative bias instability due to electron injection into polysilicon capacitors. If electron exchange of positive charge with electrons in polysilicon does not occur, positive charge generated at the polysilicon-SiO2 interface appears as fixed positive charge. For As-doped polysilicon, φms∼-0.46 V; for P-doped polysilicon, φms∼-0.52 V. These experimental values of φms are influenced by positive charge at the polysilicon-SiO 2 interface. We conclude that the Fermi level for degenerately doped n-type polysilicon is pinned close to but below the conduction band edge and does not depend on doping for carrier concentrations between 3×10 19 and 4×1020 cm-3.