Transistor Technology

[FrankieBoiledEgg]

So I've been thinking about the collapse of Moore's Law due to the inability to produce smaller transistors and I found an article of Boron Nitride nanotube transistors that use quantum tunnelling to produce a transistor without a semi conductor - without heat.

 

 “At the rate the current technology is progressing, in 10 or 20 years, they won’t be able to get any smaller,” said physicist Yoke Khin Yap of Michigan Technological University. “Also, semiconductors have another disadvantage: they waste a lot of energy in the form of heat.”

 

The article then goes on to say, 'Furthermore, there was no “leakage”: no electrons from the gold dots escaped into the insulating BNNTs, thus keeping the tunneling channel cool. In contrast, silicon is subject to leakage, which wastes energy in electronic devices and generates a lot of heat.'

http://www.mtu.edu/news/stories/2013/june/beyond-silicon-transistors-without-semiconductors.html

 

I think this could be groundbreaking if more experiments show this to be reliable and then either Intel or AMD will be brave enough to make the first leap, theoretically it could dramatically reduce prices because we'll be able to get massively higher clock speeds too.

Edited July 28, 2017 by FrankieBoiledEgg
Following Guidelines

[RileyTheFox]

4 minutes ago, FrankieBoiledEgg said:

So I've been thinking about the collapse of Moore's Law due to the inability to produce smaller transistors and I found an article of Boron Nitride nanotube transistors that use quantum tunnelling to produce a transistor without a semi conductor - without heat.

http://www.mtu.edu/news/stories/2013/june/beyond-silicon-transistors-without-semiconductors.html

Use the post guidelines for this section. Add quotes and your opinion. It's required for a tech news post. 

[Okjoek]

I hope they do figure something out. It'd suck for us to simply hit the ceiling in terms of progress in a few years because of no more transistor shrinking.

[DocSwag]

There's always going to be some heat, there is no such thing as 100% efficiency. It sounds to me like this just emits very little heat.

[S w a t s o n]

Meanwhile in 2040 we're actually still gonna be using x86-64 architecture on silicon chips because the industry never changes. "Introducing Intel's new i12-999999k based on 5nm ultra FinFet++++++ QUAD GATE Technology"

[RedRound2]

6 hours ago, DocSwag said:

There's always going to be some heat, there is no such thing as 100% efficiency. It sounds to me like this just emits very little heat.

Superconductors are 100 percent efficient at conducting electricity. Its conversion of energy that cannot be 100 percent efficient, due to thermodynamics 

[Sniperfox47]

On 7/28/2017 at 1:24 AM, FrankieBoiledEgg said:

~snip~

The issue with this is the same as the issue with a lot of other materials and processes set to replace silicon, which is to say fabrication.

 

Silicon is relatively easy to work with, it allows for very complex systems to be fab-ed with known techniques and implemented in existing plants.

 

A lot of the alternative materials not so much. They need massive investments in building new plants and new fab processes to even be viable at a much larger size than silicon transistors. That means, at least in the immediate future, they're not going to be replacing silicon on the high end side of things. Having transistors that consume 50% as much power is great, but not if you can only cram 30% of the into your chip, have higher die failure rates, and can't implement it in your upcoming product lines.

 

I don't expect to see alternative materials see very widespread use before maybe 2025. I could be wrong, but considering we haven't even really started to see them implemented in commercial products...

[Doobeedoo]

I'm excited to see a revolutionary change once we shrink and use silicon to it's limits. Chips from new type of materials and building blocks. Just imagine, way less energy needed, no actual cooling even needed down the line with significant performance increase along. Something like that will be epic.

We'll still be on silicon for a while no doubt, still can shrink it, improve architecture and so doing 3D designs and interposer stuff.

[DocSwag]

7 hours ago, RedRound2 said:

Superconductors are 100 percent efficient at conducting electricity. Its conversion of energy that cannot be 100 percent efficient, due to thermodynamics 

Even so there is going to be some loss. For example, the electrons moving are going to create a magnetic field which will induce some current somewhere else, even if it is very little. 

[RedRound2]

58 minutes ago, DocSwag said:

Even so there is going to be some loss. For example, the electrons moving are going to create a magnetic field which will induce some current somewhere else, even if it is very little. 

Current moving in a conductor doesn't directly induce electricity into another conductor. There needs to be relative movement between them iirc. You don't get free energy if you place one live wire with another dead wire.

My concept is a bit dusty, but I don't think a superconductor loses energy through magnetic induction, in fact, no conductors do but rather it's the resistance

[DocSwag]

23 minutes ago, RedRound2 said:

Current moving in a conductor doesn't directly induce electricity into another conductor. There needs to be relative movement between them iirc. You don't get free energy if you place one live wire with another dead wire.

My concept is a bit dusty, but I don't think a superconductor loses energy through magnetic induction, in fact, no conductors do but rather it's the resistance

Poop balls I did a brain fart. Of course there needs to be relative movement derp.

 

Transistors are switching though, so voltage and current is going to be changing which will cause some induction. Plus technically there will be a teeny bit of induction just from jiggling happening from heat.

[FrankieBoiledEgg]

On 29/07/2017 at 4:59 PM, DocSwag said:

Poop balls I did a brain fart. Of course there needs to be relative movement derp.

 

Transistors are switching though, so voltage and current is going to be changing which will cause some induction. Plus technically there will be a teeny bit of induction just from jiggling happening from heat.

The "Jiggling" wouldn't be enough to cause an induction, this is why extremities in calculations are left out of equations as their effect is just so minute it won't cause any real-life implications.

[mynameisjuan]

Holy necro batman. 

[S w a t s o n]

18 minutes ago, mynameisjuan said:

Holy necro batman. 

Listen, the OP just came back after almost exactly an entire year to make his second post, necro-ing his own thread which was his first post. Truly I can die peacefully now, I've seen it all

[bcredeur97]

On 7/28/2017 at 3:49 PM, Okjoek said:

I hope they do figure something out. It'd suck for us to simply hit the ceiling in terms of progress in a few years because of no more transistor shrinking.

this will be so bad for the economy... it's almost scary to think about it.

we function on the reliance of progress. no progress; everything kinda falls apart.

[DocSwag]

Why couldn't this be resurrected 3 days later so it would be a year  

[cj09beira]

4 hours ago, FrankieBoiledEgg said:

The "Jiggling" wouldn't be enough to cause an induction, this is why extremities in calculations are left out of equations as their effect is just so minute it won't cause any real-life implications.

anyway i dont think this would mean we would go to low power devices if anything they would just use this to be able to stack more dies one on top of the other without overheating the lower dies and go for performance instead.

there are other transistors being worked with too like light transistors and electron spin transistors (i dont know much about any of those though)

[Okjoek]

WTF. His second post on the forum over a year later is a reply to me?! 

 

[asus killer]

On 7/29/2017 at 3:07 AM, Swatson said:

Meanwhile in 2040 we're actually still gonna be using x86-64 architecture on silicon chips because the industry never changes. "Introducing Intel's new i12-999999k based on 5nm ultra FinFet++++++ QUAD GATE Technology"

with 2500 cores at 5Ghz. Evolution baby