Re: IBM - New SUB-Nanometer STACKED Chip

On 6/30/26 22:09, Paul wrote:
> On Tue, 6/30/2026 12:06 PM, c186282 wrote:
>> On 6/30/26 09:51, Mr. Man-wai Chang wrote:
>>> On 6/30/2026 5:14 PM, c186282 wrote:
>>>>>
>>>>> Can you fabricate 0.000000...0000000001 nm chips?
>>>>>
>>>>> Is zero the seal or wall? :)
>>>>
>>>>      Um, pretty quick you get to ATOMS ... and, for any
>>>>      normal electronics, that's IT.
>>>
>>>
>>> You cannot have 0.000000....00 nm chip.
>>>
>>> That's a void, empty, nothing. :)
>>
>>    "Electronics" are now about literal atom-thick structures.
>>    Can't go any smaller.
>>
>>    Any better future stuff will have to exploit quantum
>>    effects - get more bang for yer nanometer. Alas quantum
>>    stuff isn't as deterministic as bulk matter devices
>>    and suffer from the uncertainty principle.
> 
> But only if there are properties that actually have some use.

   True.

   There do seem to be SOME uses - but "general utility"
   is not so clear.

> A quick Google tells me there is a thing called "spintronics"
> which can make insulators. The materials are "foreign" to
> current semiconductors (so the materials may not "play nice"
> with the rest of the substrate). And having an insulator,
> does not give me a switching transistor.

   Umm, no. "SpinTronics" may have uses - but, as you said,
   it does not segway neatly with existing tech.

   (fun - my spell-checker thing wants "Segway" - the
   two-wheeled commercial thingie :-)

   There are all sorts of exotic potential tech - weird
   probability waves spreading over 2-D surfaces and such.
   Can they be made into USEFUL technologies ? Maybe 10%
   and only for specialized uses.

   Existing transistor tech - it's deterministic, solid.
   Alas with the IBM product we've probably reached the
   bitter END of conventional electronics. Some seriously
   different stuff will be needed for The Future if we
   need more speed/density.

> Historically, we've made good use of quantum mechanical "tunneling",
> which is finding a way underneath energy barriers. We have one
> afternoon lab using one of these (these diodes are used in
> time domain reflectometry for the nice sharp edges). I put mine right
> across the terminals of an HP gray-plastic power supply, the one with
> the course and fine controls. And you can use the power supply
> as a curve tracer, and walk the diode right up to the
> point where it switches. The power supply has amazing
> characteristics at DC (not so much at HF). And that was
> a fun afternoon of farting around.
> 
>    https://en.wikipedia.org/wiki/Tunnel_diode
> 
> And that's an illustration of an effect, it's not me predicting
> they'll run right out and make something out of that.
> 
> This is another example of tunneling.
> 
> https://en.wikipedia.org/wiki/Flash_memory#Fowler%E2%80%93Nordheim_tunneling
> 
>     "The Fowler-Nordheim tunneling effect is reversible, so electrons
>      can be added to or removed from the floating gate, processes
>      traditionally known as writing and erasing."
> 
> There's no uncertainty principle there, you need enough electrons
> to maintain noise immunity.


   NO longer sure if you go below 1nm alas.


> If there is some property that can be used, and is not a flaky pastry,
> then it will be introduced slowly. There is too much money involved
> for this to be just "science", it's "business" too. And that affects
> how you do it.

   Expect a LOT of flaky pastries in short order.

   One or two will be useful/do-able.

   The rest ....