(Invited) Effects of Ge Substrate Annealing in H2 on Electron Mobility as well as on Junction Leakage in n-channel Ge MOSFETs

Wednesday, October 14, 2015: 10:00
105-B (Phoenix Convention Center)
A. Toriumi, C. Lee (The University of Tokyo), and T. Nishimura (The University of Tokyo)
A significant enhancement of electron mobility and understanding of carrier transport physics in Ge MOSFETs have been made thanks to significant reduction of major extrinsic carrier scattering sources such as Ditand surface roughness, in addition to EOT scaling in Ge gate stacks. Since the phonon scattering is the dominant mechanism in Ge bulk, Ge substrate considerations have been out of scope for improving the carrier mobility in Ge MOSFETs. In this talk, however, we report that the peak electron mobility is dependent on how Ge substrates are treated.

We found that H2 annealing was quite effective for achieving the atomically flat surface of Ge substrates. In parallel with this effort, we recently found that H2annealing of Ge substrates could improve the peak electron mobility significantly, even though the peak electron mobility on as-received Ge substrates was poor. This fact indicates that there is definitely another origin of electron mobility degradation inside Ge bulk, because the peak electron mobility is not significantly affected by the surface roughness scattering. This fact indicates that electron mobility is not simply understood by considering the scattering mechanisms reported for Si MOSFETs. Here, we report the relationship between electron mobility and oxygen atom concentration in Ge substrate.

 Furthermore, we discuss effects of oxygen atoms in Ge substrate on reverse-biased n+/p junction leakage current. The oxygen atoms in Ge substrates experimentally reduced the leakage current in n+/p junctions. This result is good news for Ge technology, but it is incompatible with high electron mobility in terms of achieving high performance MOSFETs. Then, we fabricated MOSFETs with selective areal profile of oxygen atoms along the channel and source/drain regions. This method has enabled us to achieve high electron mobility in n-channel Ge MOSFETs together with high Ion/Ioffratio.

Although the junction leakage has been considered to be intrinsically poor in Ge due to a narrower energy band gap, this paper strongly suggests that Ge technology will be further improved for the best with the help of a rich heritage of Si technology.