An approach using a subamorphizing threshold dose silicon implant of optimal energy to achieve shallower junctions

Significant channeling during the ${\mathrm{B}}^{\mathrm{+}}{\mathrm{/BF}}_{\mathrm{2}}^{\mathrm{+}}$ implant and enhanced diffusion during the subsequent anneal step limit the ability of forming ultrashallow $p^{+}$ -type junctions. The problem is becoming even more serious as device dimensions are shrinking and shallower junctions are needed to prevent short channel effects. An approach using a subamorphous threshold dose ${\mathrm{Si}}^{\mathrm{+}}$ preimplant of optimal energy is proposed and demonstrated, and it reduces both channeling during the subsequent dopant implant and enhanced B diffusion during the anneal step. In this approach, the ${\mathrm{Si}}^{\mathrm{+}}$ implant is performed prior to the dopant implant ${\mathrm{(B}}^{\mathrm{+}}{\mathrm{/BF}}_{\mathrm{2}}^{\mathrm{+}}\mathrm{)}$ ; thus it is referred to as a ${\mathrm{Si}}^{\mathrm{+}}$ preimplant. The energy of the ${\mathrm{Si}}^{\mathrm{+}}$ preimplant determines the spatial distribution of the vacancies and interstitials formed by it. The vacancy excess created by the ${\mathrm{Si}}^{\mathrm{+}}$ preimplant creates strain in the lattice by disrupting the periodic potential of the lattice, which in turn helps in reducing channeling during the subsequent dopant implant. The vacancy distribution created by the ${\mathrm{Si}}^{\mathrm{+}}$ preimplant also recombines with the interstitials produced by the dopant implant resulting in fewer interstitials that can contribute to the enhanced diffusion of B, thereby reducing the enhanced B diffusion and thus resulting in shallower junctions. The energy of the ${\mathrm{Si}}^{\mathrm{+}}$ preimplant is chosen such that the excess interstitials produced by it lie beyond the depth to which the dopants would diffuse in the absence of a ${\mathrm{Si}}^{\mathrm{+}}$ preimplant, thus the interstitials produced by the ${\mathrm{Si}}^{\mathrm{+}}$ preimplant do not contribute to enhanced B diffusion. Junction depth reduction was studied using unpatterned $n$ -type wafers by secondary ion mass spectrometry analysis. A ${\mathrm{Si}}^{\mathrm{+}}$ preimplant of the desired energy (ranging from 50 to 100 keV) was then performed through a thin screen oxide. The control samples were not implanted with ${\mathrm{Si}}^{\mathrm{+}}$ . Subsequent ${\mathrm{B}}^{\mathrm{+}}$ or ${\mathrm{BF}}_{\mathrm{2}}^{\mathrm{+}}$ implants were carried out at 5 keV. The samples were annealed at 1000 $°\mathrm{C}.$ The junction depth was reduced by 200 Å for 5 keV, ${\text{6×10}}^{14}\hspace{0.5em}{\mathrm{cm}}^{\mathrm{-2}}$ ${\mathrm{BF}}_{\mathrm{2}}^{\mathrm{+}}$ implants and 450 Å for 5 keV, ${\text{2×10}}^{14}\hspace{0.5em}{\mathrm{cm}}^{\mathrm{-2}}$ ${\mathrm{B}}^{\mathrm{+}}$ implants using this approach of preimplanting with a neutral species $({\mathrm{Si}}^{\mathrm{+}}$ was used in this study) of optimal energy. Low leakage devices have been fabricated using this approach to demonstrate that the