Near-surface dopant passivation after wet-chemical preparation

Abstract

Recent advances in wet-chemical surface treatment of silicon has resulted in unreconstructed, hydrogen terminated Si(111) surfaces with extremely low surface recombination velocity of only 0.25 cm/s [E. Yablonovitch et al., Phys. Rev. Lett. 57, 249 (1986)]. This points towards a surface free of surface states that normally pin the Fermi level EF on Si(111) somewhere near midgap depending on the surface reconstruction. This expectation is not borne out, however, by experiment. Using Si 2p core level spectra, we find consistently a surface Fermi level position near midgap for 16 heavily doped n- and p-type samples. Checks for surface photovoltage and ion concentrations sufficient to induce the observed band bending were negative. All samples attain their bulk Fermi level position at the surface after annealing to about 400 °C, a temperature at which the surface hydrogen coverage is unchanged as monitored by ultraviolet excited photoemission. Above 400 °C, hydrogen starts to desorb from the surface and only with the onset of the 7×7 reconstruction does EF move back towards midgap. We suggest that the initial position of EF near midgap is due to a hydrogen induced passivation of dopants that leaves an intrinsic surface layer. Annealing at ∼400 °C reactivates the dopants by breaking the B–H and P–H complexes as is known from plasma passivation. The necessary passivation depth has been calculated as a function of bulk doping yielding values of the order of one micrometer.