When Do We Replace Si With Ge?, Higher mobility makes for better channel

Electrons and holes have higher mobility in Ge than in Si. With SiGe epitaxial technology being used already in mainstream, it doesn't seem remote at all to consider the use of Ge as the semiconductor channel layer, thereby (in combination with metal gates and dielectrics not containing silicon) enabling the complete removal of silicon (perhaps even Si wafers) in future CMOS technology.

I assumed Ge wasn't used because of the higher leakage current (we're going back over 30 years now). I thought that might be a consequence of the lower band gap but could well be wrong.
Is the leakage problem 'solved'? Or maybe not a problem in MOG (!) technology?
-C2.

I wouldn't like to go back to germanium!
Its maximum operating temperature is too low, due to the small gap.
I've never heard of a good interface with any insulator, however this is a vital need.
Breakdown field is also smaller, due partly to the smaller gap, however this is also a vital need for small circuits.

Hole mobility would be better, but only the surface mobility is interesting for Cmos, and this requires a good interface to insulator.
Yes, SiGe is used very locally on bipolar transistors. Not for the mobility, but to get a smaller gap locally, thus injecting quick ("hot") electrons.

Try it rather with diamond or III-V compounds. But remember the silicon industry invests billions in silicon, so your competitor semiconductor needs to be damned better just to keep pace. 30 years ago, I used GaAs to make faster chips, but silicon Cmos has crashed everything meanwhile.

The junction leakage can be reduce to ~ picoamp/um^2 with annealing. Or you can use GeOI.

BTB is also significant but can be addressed through low-voltage device structure design.