Search

Fin field-effect transistor (FinFET) scaling beyond the 10-nm node requires formation of a junction isolation region between the source and the drain to suppress sub-fin leakage current. In this paper, heavy species such as Sb and As were implanted at room temperature to form a punch-through stop (PTS) layer in ${n}$ -Ge substrates. The impact of PTS implants on channel doping and defects, as well as junction leakage, was investigated for bulk Ge ${p}$ -FinFET applications. We report retrograde doping profiles with a low surface (channel) dopant concentration of 4–5 $\times 10^{\textsf {16}}$ cm−3 for Sb PTS implants post 550 °C and 650 °C activation anneals. Besides achieving low effective channel doping, the 650 °C high-temperature activation anneal post PTS implant also helps in the reduction of channel defect density. Furthermore, low junction leakage currents are also demonstrated for Sb and As PTS implants with 550 °C and 650 °C activation anneals. To avoid damaging the FinFET channel layer due to PTS ion implantation, Ge homoepitaxy was also studied post Sb PTS implants. However, Sb PTS implants activated at 650 °C without Ge epitaxy exhibit the best results in terms of low channel dopant and defect concentrations, besides junction leakage. [ABSTRACT FROM AUTHOR]

The density and origin of dislocations in a variety of molecular-beam-epitaxy-(MBE)-grown Ge 1−x Sn x /Ge/Si (0 0 1) heterostructures, with Sn-content varying from 0.4 to 5.2%, have been investigated systematically by high resolution X-ray diffraction (HRXRD). In particular, using the approach due to Kaganer et al. (V.M. Kaganer et al., Phys. Rev. B 72, 045423 (2005)) for the first time to the Ge 1−x Sn x alloy, it is demonstrated that reliable estimates of both edge and screw dislocation densities can be obtained from HRXRD data. Based on the correlations of strain relaxation and dislocation densities of the alloy epilayers and the underlying Ge buffer layers, we observe that dislocations threading from the latter predominantly contribute to the strain relaxation of the former. Thus, Ge 1−x Sn x epilayers of sub-critical thicknesses can be made to relax significantly by growing them on partially-relaxed, relatively-thin Ge buffer layers. This may be promising for the realization of Ge 1−x Sn x epilayers with direct electronic band gap. [ABSTRACT FROM AUTHOR]