It's probably very hard to shrink further than a very low number of atoms but we're far from that. The "2nm" marketing name has no relation to the actual physical size of the transistors. The physical features of the "3nm" process node are something like 24nm and 48nm. The Van der Waals radius of a silicon atom is about 0.2nm. Transistors made from a single molecule have been demostrated. Maybe we could even shrink to elementary particles but that's in the realm of SciFi so far.
The absolute hard minimum has to be one silicon atom wide, or 0.2nm. But note that "3nm" doesn't quite have any features that are that size, it's a non-precise marketing term.
Do we know what size these structures really are?
I dont know about 3nm but for 5nm they still have feature size ranging from 5x nm to 2x nm.
That would be surprisingly accurate
Yes, in addition to the limits set by atomic radius, you also have the issue of electrons tunneling through the tiny energy barriers that exist when the the transistors are so close together.
The smallest transistor is bounded by the atomic radius; a recent paper boasts 0.34nm for a graphene transistor. Scaling that up, to borrow a phrase from quantum computing, is "just an engineering problem." But for now we can only produce things on the atomic scale a handful of atoms at a time, and making a trillion-transistor processor would take somewhere between thousands and billions of years with today's tech.
So far any prediction of the bottom has not really come true. There must be some limit, but these engineers are pretty crafty. I bet they won’t beat one transistor per atom! Any other prediction, I’m suspicious about.
Physically impossible? I wouldn't say that. 2nm is a marketing term, depending on how you define it, it is actually 20nm or 45nm here.