Back when I was a kid and first learned about Peltier coolers, I mistakenly thought they had the ability to transfer heat remotely from one surface to another via wires. I didn't realize the heat transfer took place within a set of PN junctions in a substrate, and basically just moves heat from one side of the material to the other.

I was wondering if the concept of "transferring heat to a remote surface via wires" is valid? In other words, like a split thermoelectric cooler with the cold side at one location and the hot side at another location, connected by wires, using electrical current to "push/pull" heat from one place to another

Or is this a better question for the physics site?

Or is this just stupid? :)

  • 1
    \$\begingroup\$ Well... it might be stupid. Look at heat pipes for an alternate solution. \$\endgroup\$
    – Samuel
    Jun 9, 2015 at 21:45
  • \$\begingroup\$ I'm familiar with heat pipes, but was wondering if heat could be "pushed" via applying current to create an even greater transfer? (edited question for clarity) \$\endgroup\$ Jun 9, 2015 at 21:53
  • \$\begingroup\$ generating a thermal difference between one side of something which is thermally conductive and another will encourage heat to equalize towards the colder area, using the conducive material as a 'medium' through which to flow. Cool down some other area, and it will "pull" heat towards it from the other area. The rate of thermal diffusion is proportional to the difference in temperature \$\endgroup\$
    – KyranF
    Jun 9, 2015 at 22:07

3 Answers 3


This isn't going to work.

Metal is an excellent conductor of heat, so its temperature quickly becomes the same throughout. The figure of merit for a thermoelectric device is crippled by high heat conductivity [\$\kappa\$].

$$ZT = {S^2GT \over \kappa}$$

Additionally, copper has a Seebeck coefficient [P] of 1.5 μV/K. The most common material used, bismuth telluride, has a coefficient of -297 μV/K. Notice this is the squared term in the numerator, it's very relevant. This is why peltier devices are not made of metal. Increasing the current will only waste more energy.

Result, you get a heater.

Ok, you say, I want my wire to be made of the same material inside a typical junction.

The problem there is the thermoelectric effect has a limited efficient distance for any material. As the thickness of the junction material increases the electron mean free path doesn't. Instead of heat transfer you get heat generation (see Fourier's Law).

Result, you make a heater.

Ok, you say, I will make many small stacks of peltier junctions in my wire passing heat along like a bucket brigade.

The problem here is each device generates its own heat while trying to pass along heat from cold to hot sides. Stacking devices quickly becomes very inefficient and things get hot very quickly.

Result, you make a heater.

  • \$\begingroup\$ Thanks Samuel. Again to clarify, I'm not trying to build anything. I'm just trying to determine if it's theoretically possible to utilize electrical current to move heat from one place to another, the 'places' being some sort of devices connected by standard wires. \$\endgroup\$ Jun 10, 2015 at 5:14

I think your old recollection is correct.

A thermocouple sensor has a hot end and a cold end, and generates a voltage proportional to the temperature difference between the ends.

Seebeck discovered that a current would flow when wires of two dissimilar metals were joined at the ends, and one junction made hotter than the other.

It's clear that in this experiment, the wires could be made as long as you like; the temperature of the junctions is what matters. And if current is flowing and work being done, then heat is being transferred from one end to the other.

Peltier plates are wide and flat because they are designed to be more efficient than simply twisted wires, they have large area junctions. But the effect is the same,it's at the hot and cold junctions between materials that the heating and cooling takes place. Any heat conduction through the Peltier device is an unwanted effect, reducing its performance.


To extend on the comment about heat-pipes and add in some answering-powder, real quick before I go back to work:

If you put a Peltier element on something, then connect a heat-piping mechanism to that, then connect another Peltier element at the other side of the heat pipe, you will be "pumping much faster".

Any system or material that transports heat does that somewhat reluctantly (some extreme and currently real-world-infeasible examples excluded).

This reluctance can be expressed as a thermal resistance. The flow of thermal power depends on the thermal resistance and on the thermal potential in that resistance, just like in electronics.

So to make the heat flow faster, you can decrease the resistance:

  • Increase the flow of oil or water in a coolant system
  • Make a metal heat-pipe thicker
  • Blow on the object in the case of air-transfer.


But you can also keep the resistance the same and increase the potential, this "potantial" is expressed in a thermal difference. So if one end of the heat-pipe (use that for the example) is 50degrees C, and the other is 30degrees C, the differential is 20 degrees and the heat will flow with speed x.

If you now heat up one end of the heat-pipe to 70degrees C and cool down the other to 10degrees C, you have a differential of 60 degrees and if the resistance stays the same (real-world: not exactly, but, in a margin...) your thermal energy will be able to flow at 3x speed.

So, if you have a device that is 50 degrees C, and an environment of 30 degrees C, but need the heat to flow through a heat-pipe between them as quick as if there's a difference of 60 degrees C, then you can put a peltier on each end, capable of a thermal differential of 20degrees C under the circumstances.

Other than that, another answer would have been (as was already hinted at above):

Just use a pump, some hoses and a decent type of oil: Done! :-)

EDIT: Due to comment-y adjustment of the parameters, more options:

You asked/clarified in a comment: "Yes, but what if, temperature goes through a wire as electrons and comes out as temperature on the other side?", right?

Well, that is both possible and not possible. It certainly isn't feasible in any big way, but it is actually possible to test that in the real world. You will just be wasting a shit-load of energy until you Eureka yourself a better system that nobody's though of before with much higher efficiency. Though I'd be very surprised if you get actual mentionable numbers for the efficiency. But I'm happy to be surprised most days, so have at it.

What I'm getting at is the following fun fact:

A Peltier element doesn't just work one way. You can also use it to generate electricity from the thermal differential. If you apply boiling water to one side, and run room-temperature water along the other, that will induce a voltage. Its maximum current and voltage won't be super impressive, but they're there.

So, for the thought experiment, imagine you have the Peltier element neatly connected to the ground, which is mega-thermally conductive in your back-yard and its temperature is -20 degrees Celcius, because it's winter. Or you live somewhere else than you think you do, whichever works for you, just imagine it.

So, that side is always -20 degrees C, because the ground will take away anything you throw at it. Yet, your feet are warm. And also made of thermally super-conducting metal - I mean, I'm already claiming your backyard is a thermal-super-conductor at some fixed temperature, so I might as well just claim that as well. Oh and if I'm simplifying a bit further, your body generates heat infinitely fast if your foot-temperature drops a 1/100th of a degree.

((Small side note: Imagine having to accept all that ^^. .... That's how newly become professionals feel when people simplify op-amp or transistor characteristics. ;-) ))

Done imagining? Good.

Now, imaginarily step onto the Peltier. Hurray! Voltage!

You can step off now, since nothing is getting cooler or warmer, we can easily get our ever-ready spool of super conducting AWG30 wire from the garage, all the parameters will be just the same in a little while.

Now, obviously, while at the garage, don't forget to look into the never-ending box of useful stuff and get out another Peltier element. There's infinitely many in there anyway.

Hook up the second Peltier element to the first one, walk to your front yard, to prove that it works you always need an imaginary house in between set-ups. It's a well known fact.

Put down the element there, such that its cold side is on the ground, which there also happens to be thermal-super-conductor at exactly -20 degrees C.

Walk back.

Now the fun stuff really starts: Because this is a thought experiment we can also just say that a Peltier Element is always 100% efficient, regardless of how you use it and it will magically be so. Let's also say that the element can wick away a couple of kilowatts of energy at the temperature differential of 37 vs -20 = 57 degrees C.

Now we have made ludicrous assumptions -which we are allowed to- about everything in this experiment, except for the basic functionality of the inner-guts of the Peltier element. And that's what it's about.

So, now step onto the element in the back yard: Poof! The snow in the front yard just started melting! QED.

Oh, did I forget to tell you? While you were walking back it snowed a meter thick exclusively in your front yard.

Exiting the imaginary world that I nearly permanently host on the inside of my head

Of course my little magic world inside my head, and hopefully yours a little now too, doesn't mean that much in the way of a promise of functionality. But it does help imagine the theory behind it.

When you stepped on the plate in the backyard, electrons were carried along the junctions because of the forced flow of heat. Those electrons shuffled along our wire to the other one and carried energy out from the earth into the plate, radiating it into my layer of super-fast-falling snow.

Obviously, if you did this even with a large peltier element and both feet, you'd be lucky to get any humanly detectable differential at all on the other element.

So, inside my head: Yup, your idea works. For now, unfortunately only there. Or at least, feasibly.

For a fun look at some real-world stuff youtube "thermo-electric generator", it'll probably get you a lot of steam-powered stuff, but I remember a few years back seeing some videos of people using TECs (thermo-electric coolers / peltier's) to make a small fan spin with boiling water versus tap water in the winter and things like that.

At the time I was thinking about a tongue-in-cheek gizmo that charged a cap from thermo energy that then pulsed its power through a diac/triac into a 230V pilot valve closing the cold water tap, but I never actually did any maths to see how much fresh water I'd be wasting before the tap got turned off.

  • \$\begingroup\$ Yes I realize there are reasonably efficient ways of carrying heat from one place to another, but I was just looking for theory on whether heat could be transferred through standard wires from one "device" to another, using electrical current as the "pump". \$\endgroup\$ Jun 10, 2015 at 20:44
  • \$\begingroup\$ @RyanGriggs yes and no. Almost the same explanation too. I'll edit an update in a bit. \$\endgroup\$
    – Asmyldof
    Jun 10, 2015 at 20:48
  • \$\begingroup\$ I do understand that heat will flow through wires as they are thermal conductors. I'm trying to think in terms of the Peltier-type heat transfer where heat is "pushed" or "pulled" to one side/end of a thing, due to the electrical current flowing through it. \$\endgroup\$ Jun 10, 2015 at 20:54
  • \$\begingroup\$ @RyanGriggs I understand. That's what my comment tried to answer. I'm just busy on this: electronics.stackexchange.com/a/174703/53769 \$\endgroup\$
    – Asmyldof
    Jun 10, 2015 at 21:35
  • \$\begingroup\$ @RyanGriggs did my edit understand your comment correctly? If so, prepare to depart for a magical land inside my head. \$\endgroup\$
    – Asmyldof
    Jun 11, 2015 at 0:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.