posted by
martyb
on Tuesday February 19 2019, @08:27PM
from the does-it-also-make-the-room-darker? dept.
from the does-it-also-make-the-room-darker? dept.
Researchers at the University of Michigan ran a light emitting diode (LED) with electrodes reversed in order to cool another device mere nanometers away. The approach could lead to new solid-state cooling technology for future microprocessors, which will have so many transistors packed into a small space that current methods can’t remove heat quickly enough.
This could turn out to be important for future smartphones and other computers. With more computing power in smaller and smaller devices, removing the heat from the microprocessor is beginning to limit how much power can be squeezed into a given space.
https://www.rtoz.org/2019/02/18/running-an-led-in-reverse-could-cool-future-computers/
[How does this compare to a Peltier device?
--Ed.]
This discussion has been archived.
No new comments can be posted.
Running an LED in Reverse Could Cool Future Computers
|
Log In/Create an Account
| Top
| 47 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
(1)
(Score: 4, Interesting) by Taibhsear on Tuesday February 19 2019, @08:32PM (23 children)
I'm not an electrical engineer but I thought the whole point of a diode was that it can only run in one direction... What kind of star trek reverse polarity shenanigans is going on here?
(Score: 2, Insightful) by Anonymous Coward on Tuesday February 19 2019, @08:41PM
I was thinking the same thing and the article is absolutely no help here.
If I had to guess, I would say that certain diodes that emit light, emit electricity when bombarded by infrared light, i.e. thermal emissions. If you reverse their polarity then the current generated is pumped back to ground thereby providing a pathway of heat return to the power source.
This is literally, only a guess and I've never heard of a substance that both emits light when powered and emits power when lighted. It would be like plugging your solar cell into a battery and having it glow. I guess it might radiate thermally though, so who knows?
We tend to think of light and heat as different things, but they are the same thing just at different frequencies. I've seen and used sound transducers, but a light transducing substance though, to my knowledge does not exist.
(Score: 0) by Anonymous Coward on Tuesday February 19 2019, @08:43PM (2 children)
You are correct. The LED will be in high resistance mode until a reverse breakdown voltage. Then it conducts.
The LED has a tiny crystal deep inside a heat insulating package. Until I see an article that explains the effect in terms of physics, I will consider this article to be a fake. (The link currently returns 403 Forbidden.)
(Score: 4, Informative) by takyon on Tuesday February 19 2019, @09:17PM
This is the real TFA and associated science article:
https://news.umich.edu/running-an-led-in-reverse-could-cool-future-computers/ [umich.edu]
http://dx.doi.org/10.1038/s41586-019-0918-8 [doi.org] (says DOI not found right now)
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Tuesday February 19 2019, @09:24PM
nonsense.. if you do read the actual article then you'll see that they have found a way of sucking up the infrared-heat-radiating-photons -- taking them (their energy) out of the system via the diode. filed that one under freaking-awesome.
(Score: 2) by krishnoid on Tuesday February 19 2019, @08:48PM (5 children)
Now no longer a light emitter, it's now a light sucking diode, with its concomitant psychedelic effects.
(Score: 2) by PartTimeZombie on Tuesday February 19 2019, @09:07PM (4 children)
It still emits, but it is now emitting dark instead of light.
(Score: 1) by fustakrakich on Tuesday February 19 2019, @09:15PM (3 children)
Does it make the room darker?
La politica e i criminali sono la stessa cosa..
(Score: 2) by DannyB on Tuesday February 19 2019, @09:20PM
Not if the room is already at absolute dark.
I'm now going to run a resistor in reverse polarity to increase current flow in the circuit.
What happens to the temperature if I put the reverse polarity LED in series with the reverse polarity resistor?
The lower I set my standards the more accomplishments I have.
(Score: 2) by PartTimeZombie on Tuesday February 19 2019, @10:08PM (1 child)
Yes, unless the room is at absolute dark, as DannyB has pointed out. (Obviously).
(Score: 1) by fustakrakich on Tuesday February 19 2019, @11:26PM
You gotta have pretty good night vision to go around hooking up LEDs and resisters in the dark, don't you?
La politica e i criminali sono la stessa cosa..
(Score: 4, Informative) by noelhenson on Tuesday February 19 2019, @09:09PM (4 children)
In a reversed polarity circuit, an LED will generate a photo-voltaic current when exposed to light. In older LED displays, unused LED segments were sometimes used as light sensors to control the display brightness. For IR remote controls, I have used the IR LED as the IR sensor as well.
I believe it could be this effect that carries away the heat. I still don't think it could carry away enough heat to make any real difference.
(Score: 2) by Snow on Tuesday February 19 2019, @09:20PM (2 children)
Interesting.
So essentially it's a solar panel, but tuned for infrared (heat)?
The article is scant on details, but I was picturing something similar to the Peltier effect where the heat is moved to another location. In this case, the heat is not moved, but rather converted to electricity?
(Score: 0) by Anonymous Coward on Tuesday February 19 2019, @09:27PM (1 child)
yes. the photon's charge gets carried out as electrons. this freaking awesome insight could just make the way for viable 3D computing structures in which waste heat converts back to charge.
(Score: 2) by qzm on Wednesday February 20 2019, @10:57AM
if only efficiency didnt matter..
(Score: 3, Informative) by jmorris on Wednesday February 20 2019, @03:40AM
Yup, many forget or never realized the truth of diodes.
All PN junctions are diodes.
All PN junctions are zener diodes, although sometimes they are destroyed at about the same reverse voltage. Just depends on the way it was designed.
All PN junctions are LEDs at some wavelength, often infrared.
All PN junctions are photodiodes, sensitive to some wavelengths more than others.
All PN junctions are solar cells, usually very poor ones because so small but still...
The reactions to light are often eliminated by encapsulating them in an opaque substance such as epoxy.
(Score: 4, Funny) by Thexalon on Tuesday February 19 2019, @10:37PM (4 children)
It's been a while since I've played with this stuff, but I recall running LEDs the wrong way on DC circuits at least would tend to turn them into "Smoke Emitting Diodes".
The only thing that stops a bad guy with a compiler is a good guy with a compiler.
(Score: 0) by Anonymous Coward on Tuesday February 19 2019, @10:52PM
Probably a much lower voltage required, or possibly no voltage since it becomes a source?
(Score: 3, Informative) by Veyrdite on Wednesday February 20 2019, @02:00AM
This will only happen if you're not limiting power to them. Backwards an LED will block until a certain threshold voltage, then it will start conducting, just like any diode.
Source: I regularly run LEDs backwards because I'm lazy in my designs and/or clumsy in my assembly.
(Score: 4, Funny) by hamsterdan on Wednesday February 20 2019, @03:27AM (1 child)
Well, all diodes emit light at least once
(Score: 2) by DannyB on Wednesday February 20 2019, @03:23PM
Also true of Light Emitting Resistors too!
As a teen, in about 1976 ish, I put a red LED across a small rectangular 9V battery for fun. Easy to do. Just bend the LED leads a bit. Hold it with needle nose, and touch it to the battery posts. Boy this is going to be very bright but won't last long.
POOF! It exploded faster than I could have imagined. The first thing I realized was the loud POP. Then I also realized I had heard the cap of the LED hit the ceiling. Then I realized I could smell it. It all happened so fast. Like pulling the trigger on a gun.
The lower I set my standards the more accomplishments I have.
(Score: 2) by c0lo on Tuesday February 19 2019, @10:37PM
Nobody is perfect, see the leakage current of a diode [learningaboutelectronics.com]
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 0) by Anonymous Coward on Wednesday February 20 2019, @06:38AM
Spock is now a swash-buckling chick-magnet with a big pointy dick, and Kirk plays 3D chess all day.
(Score: 2) by DannyB on Wednesday February 20 2019, @03:27PM
Data says it is nothing compared to when you subject the reverse polarity LED to an inverse vertiron particle displacement field. Spock adds that it works better with a hiesenberg compensator.
The lower I set my standards the more accomplishments I have.
(Score: 5, Interesting) by JoeMerchant on Tuesday February 19 2019, @09:07PM (14 children)
While the LED-heat-pump might relocate heat away from a critical area, it will require a net increase of power dissipation on the die to use this kind of heat-pump scheme.
The heat goes somewhere - hopefully they can use these heat-pumps to more efficiently transfer heat from inner layers of 3D chip-stacks out to the environment (heat sinks).
Still, if your processor nominally consumes 15W but has localized overheating problems, you might expect to consume 20+W while using the heat pump tech to relocate the heat away from overheating areas.
Would be a fun thing to develop - placing temperature measuring thermocouples in hundreds of places throughout the die, running an intelligent heat distribution algorithm to keep the hot spots in-spec, but not running more heat-pump than necessary...
🌻🌻 [google.com]
(Score: 5, Informative) by Immerman on Tuesday February 19 2019, @09:27PM (10 children)
It's unlikely to function as heat-pump though. More likely the light-emitting junction, in reverse polarity, becomes a photon-absorbing junction, converting infrared photons to electricity rather than the other way around.
Basically, it transforms thermal radiation into electricity, it doesn't have anything to do with mechanical heat, and thus is not bound by the usual thermodynamic limitations.
(Score: 4, Touché) by Anonymous Coward on Tuesday February 19 2019, @10:05PM (7 children)
The great thing about being able to convert heat into electricity with no thermodynamic losses is that you can apply some electric power, let the power dissipate to heat, convert the heat back into electricity and feed it back into the power supply. Then you disconnect the 'bootstrap' power, and let your perpetual motion machine run forever mining bitcoin.
Another useful feature of direct conversion of heat into electricity is that you can make an air con unit that not only cools your house, but also powers the rest of your house. You could even make money by feeding it back to the grid.
(Score: 2) by c0lo on Tuesday February 19 2019, @10:24PM (6 children)
Even more, you do it in perpetuity.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by JoeMerchant on Tuesday February 19 2019, @10:58PM (1 child)
Unfortunately, free market commercial demands for cheap energy means that we'll all freeze to death with all this heat converted to electricity...
I think the "infrared photon converted directly to electricity" is missing something, somewhere - probably one of those laws of thermodynamics having to do with entropy...
🌻🌻 [google.com]
(Score: 3, Informative) by c0lo on Tuesday February 19 2019, @11:27PM
1. if you want to cool the junction faster than the heat diffusion, you can but you'll need to "evacuate" the electrons promoted (by IR absorption) into the conduction band before they recombine and give you the absorbed heat back. Which means you'll need to maintain the reverse polarity on the diode, and this doesn't energetically come for free. The good thing in this 'thermal pump' setup - you are going to produce the extra heat in some other place, where you may be able to dissipate easier.
2. if you think, a photoelectric element is doing the light to electrical energy conversion without applying a reverse potential (and IR is light). But in this case, the junction is highly doped and the produced electrons/holes move under the gradient of charge concentration by thermal diffusion. If you make the junction thin enough, more charge carriers will reach the collecting electrodes (and "find it easier to travel" to the opposite side of the junction using the external circuit) than they recombine and you will have an electric current. However, you can't use this arrangement to cool (i.e. extract energy) faster than thermal diffusion - because the thermal diffusion of charge carriers that is the very engine which powers the PV element.
Now, the consequence of the above: you need the junction to be "hotter" than the outside circuit, otherwise the PV element will never generate electrical power (will simply act as a resistor).
In both cases, the laws of entropy are already at work.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 3, Funny) by pipedwho on Tuesday February 19 2019, @11:40PM (3 children)
I wanted to add more, but as my house started to runaway towards zero Kelvin, my fingers were freezing and slowing down too much to continue. I was posting from my phone, and I can’t be arsed porting my password manager across, so posted as AC in my ice cold A/C, while feeding AC back to the grid.
(Score: 2) by c0lo on Wednesday February 20 2019, @02:24AM (2 children)
Sucks to be so greedy, ain't it?
If you would only export half of it and use the other half to heat your home, you wouldn't become super-conductive.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by pipedwho on Wednesday February 20 2019, @02:53AM (1 child)
Not greedy as such. Just forgot the thermostat and without the laws of thermodynamics to asympote it towards an energy equilibrium it got out of hand. Barely made it to the off switch before I was too frozen to move.
(Score: 1) by anubi on Wednesday February 20 2019, @03:52AM
Carnot is a bitch when it comes any useful energy from heat differentials.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 4, Informative) by c0lo on Tuesday February 19 2019, @10:22PM
That's exactly how it functions. Otherwise you'd be able to extract the energy from a low temperature (say, 60C) and use it to heat water to boiling point without consuming extra work.
Don't forget that the difference in electrical potential (otherwise know as voltage) is defined as 'the work necessary to transport a unit of electrical charge between the two potentials'.
To evacuate the electron with higher energy resulted from the absorption of the photon, you need to maintain the reverse voltage applied on the diode - otherwise the electron will recombine with the hole it left behind. So, having displaced the electron, as soon as it reaches the collecting electrode and neutralizes some charge there, you'll have to put the extra work to restore the voltage. And, believe it or not, you aren't going to obtain a unitary efficiency in restoring that voltage.
Nothing escapes thermodynamic laws (which are just the macroscopic average expression of 'the energy always flow from high towards low".
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by JoeMerchant on Tuesday February 19 2019, @11:03PM
Last time I looked into it, everything is bound by the usual thermodynamic limitations. You can shuffle things around, but once entropy has increased there's no decreasing it in the big picture. Entropy can be decreased in a localized part of a system, but only by increasing entropy (usually including some bonus entropy, but at least as much as was decreased) somewhere else in the system.
Unless Maxwell's Demon is real...
🌻🌻 [google.com]
(Score: 0) by fakefuck39 on Tuesday February 19 2019, @09:29PM
Maybe I'm not understanding how this works. From what I get, IR hits the LED, the LED produces moving current. The current I would presume go over a wire and power something, like a fan or be wired to an electric heating coil in a cooler area. This does increase the power output of the area, but that increase is electric, and moves offsite on a wire.
(Score: 0) by Anonymous Coward on Tuesday February 19 2019, @09:35PM (1 child)
i'm not sure that your "net increase of power dissipation" is a thing here -- since the ir-heat-photon-energy flows right out of the system. if you found something to indicates that this sort of draining would really add a 25% heat dissipation tax to the system, please cite! i'd expect the 'receiver' to be of very high resistance and 'the tax' to be very very low, perhaps even neutral to negative.
(Score: 4, Touché) by c0lo on Tuesday February 19 2019, @10:34PM
Right! So the higher the resistance the better, eh?
Geni(t)us idea: let's make it infinite by not connecting the diode terminals and we may even get a perpetuum mobile. Even better, with one less electronic component, thus cheaper.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 0) by Anonymous Coward on Tuesday February 19 2019, @10:24PM
I KNEW it would happen some day!
Freeze to 0 Kelvin in seconds!
(Score: 3, Informative) by c0lo on Tuesday February 19 2019, @11:34PM (1 child)
Such a diode is a Peltier element embedded/interleaved with the active components of the IC.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 0) by Anonymous Coward on Wednesday February 20 2019, @02:47AM
It would have an interesting effect for sure.
https://www.youtube.com/watch?v=6WGKz2sUa0w [youtube.com]
(Score: 1, Funny) by Anonymous Coward on Wednesday February 20 2019, @02:22AM (2 children)
At what point did this recently become an issue?
(Score: 2) by DannyB on Wednesday February 20 2019, @03:30PM (1 child)
I have an idea!
As a way of removing heat from a microprocessor, how about we remove transistors from a microprocessor! x86 must DIE DIE DIE !!!
No need to support old 8 and 16 bit instructions with segment registers and ability to boot DOS 1.0. Forty years of legacy baggage is time to go away now thank you.
The lower I set my standards the more accomplishments I have.
(Score: 0) by Anonymous Coward on Wednesday February 20 2019, @04:39PM
dude, I have a whole bunch of old x86 dies. I think they are called CPUs.
(Score: 2) by SemperOSS on Wednesday February 20 2019, @11:45AM (1 child)
The article is not saying anything about where the energy goes, but as conservation of energy is still a given, there are only a few possibilities:
You guess is as good as mine.
I don't need a signature to draw attention to myself.
Maybe I should add a sarcasm warning now and again?
(Score: 0) by Anonymous Coward on Wednesday February 20 2019, @05:50PM
you could make "IR absorbing" solar panels then use the generated electricity to make a "nasty hot photon"
(a laser) beam and point it into deep space.
the aliens would probably complain to the inter-galactic department responsible for managing un-natural light contamination and would
send a delegation to suggest a compromise by pointing the beam at our sun.
the project leader of "earth-icebox" would probably complain but after doing the math would realize that even though the added beam to the suns core would heat up the sun and increase its (de)evolution it would also increase the in-falling IR which is powering the beam anyways.
the sticking point is then that the sun, being huge would radiate naturally in all directions thus, the un-natural light contamination would be moot ...