Super ‘superlattices’ could enable ‘superfast’ transistors | Hacker News Reader

Super ‘superlattices’ could enable ‘superfast’ transistors

newelectronics.co.uk

77 pointsrbanffy13 comments

> Such superlattices __are believed to__ hold the prospect of enabling improved and new classes of electronic and optoelectronic devices, with applications including ‘superfast and ultra-efficient semiconductors’ for transistors.

Not to take a way from the accomplishment by the UCLA team. But I know when I see the good old research funding motivation/validation rethoric. The same phrasings are used, e.g., by the "topological insulator" community. Before that by the Graphene community ("lossless electron transport, super-fast transistor switching") for over a decade now and even earlier, by the quantum dot community ("single electron transistor").

This lead to next to nothing in term of actual consumer devices or improved performance of existing technology.

Not shilling for any "revolutionary" ideas. Just look at the history of the transistor, sometimes these developments and implementations take longer than what's convenient from a users perspective. And one could argue the difference now too is work done in the first half of the 20th century was relatively low hanging fruit compared to the limits we're trying to push now. It was a good twenty years between the first patent for the transistor and a demonstrable transistor, and then another seven years on top of that before a commercial implementation.

Ten years is an arbitrary length of time. And these things might lead to nothing, sure. But...

1. There's a long road between discovery and implementation and no reliable length of time that can always be pointed to say "that's how long it takes".

2. Road blocks and expense in implementation. Some things are just hard and depend on yet to be discovered or perfected technology to make them viable.

3. Existing technologies and processes are still sufficient, mature and cost effective. So we'll use those until the new technology can do it better and cheaper. We probably won't be using x86 and current technologies forever.

I like to think of an example I probably got from a Carl Sagan book that deals more with why throwing money at a problem doesn't guarantee results, but it was an argument for funding all science and discovery because we can't predict what discoveries will yield. It went something like: The British empire in 1850 could have spent every penny it had to develop television and it would have failed because it depended on science discoveries yet to be made. It doesn't mean television wasn't viable or that it was impossible. And no one could have predicted that some of the things discovered, which probably seemed inconsequential to the general public and a waste of time and money would help lead to television decades later.

Peter Thiel has a strong critique of this where he asserts the opposite: that both governments and science research spending today is too indeterminate.

Governments don't believe in the capacity to make big plans any more, and it's made it impossible for us to achieve big goals in the last 40 years on the scale that we used to (like with the moon landing).

Instead... failures are excused with portfolio theory, hundreds of universities end up doing hundreds of different things, with nobody really understanding the meaning or connection between any of it. Grant funding is based more on politics than on merit, and this slows down science, because few people are simultaneously both good scientists and good politicians.

https://www.youtube.com/watch?v=iZM_JmZdqCw

Personally, I'm waiting for Lattice 64.

By "superlattice" are they referring to van der Waals heterostructures?

No. You call the regular configuration of atoms of a particular material like GaAs a lattice.

Now when you start to grow one type of material on top of another in a regular fashion like layers in a sandwich:

10 nm GaAs 5 nm InAs 10 nm GaAs 5 nm InAs

You get what is called a super lattice. It is not an atomistic lattice, but a lattice of material type A and B.

so the difference is that super latices are thicker?

Got it, so they've figured out how to deal with lattice constant mismatches?

Yes they did. That's why you will mostly see combinations of binary materials like GaAs, InAs, AlAs to form super lattices.

So, what's the catch?

I really 'like' the title,

It's 'super.'

Obligatory reference to ice-nine: https://en.wikipedia.org/wiki/Ice-nine

We're almost there! EDIT: Oh wait, it does exist: https://en.wikipedia.org/wiki/Ice_IX

Reply on news.ycombinator.com