December 05, 2006
Thermal Rectifier Controls Direction Of Heat Flow
MIT's Technology Review reports on an experimental thermal rectifier made from nanotubes that preferentially allows heat to flow more easily in one direction.
Scientists have been precisely controlling electric current for decades, building diodes and transistors that shuttle electrons around and make computers and cell phones work. But similarly controlling the flow of heat in solids stayed in the realm of theoretical physics--until now.
Alex Zettl and his colleagues at the University of California, Berkeley (UC Berkeley), have shown that it is possible to make a thermal rectifier, a device that directs the flow of heat, with nanotubes. If made practical, the rectifier, which the researchers described in last week's Science, could be used to manage the overheating of microelectronic devices and to help create energy-efficient buildings, and it could even lead to new ways of computing with heat.
The difference in the ease of heat flow they produced was not enormous. But it was a good start. Think of it as a material that has a higher insulation rating in one direction than another.
Imagine two sides of a wall. Sometimes the outside is hotter. Sometimes the inside is hotter. It is not hard to imagine scenarios where you would want heat to flow out when it is too hot but to never flow in. Or perhaps you'd want to control the direction of heat flow depending on the season. The ability to easily flip around a section of thermal rectifier wall material would come in very handy.
I haven't read the article in question, but a true "thermal rectifier" would seem to violate the second law of thermodynamics. An electrical rectifier is not a problem, but a "thermal rectifier" would, like Maxwell's daemon, permit the construction of perpetual motion machines. I would bet that either the invention is not what you describe (perhaps it allows bidirectional flow but only along certain axes and is insulating along other axes), or it is not possible.
I will note that it is easily possible to construct a device that expends energy to move heat in one direction only (such as an air conditioner or heat pump), but such a device quite clearly does not violate the second law.
I thought the 'daemon' would show up. And I think Perry is right. Not my field but I'm surprised there have been no other comments.
It doesn't violate the 2nd law of thermodynamics any more than passive diode components violate Kirchoff's circuit laws.
Macro-scale devices, such as thermosyphons, already exist and are already used in industry that effectively act as thermal diodes that require no pumps, nor exotic physics to operate. There are many other contraptions that can do the same thing. One of the more interesting I've heard of is a vapor chamber heat shink that is ridged to use both surface tension of water and the leidenfrost effect to "pump" heat away from silicon die IC's, greating improving the effectiveness of heat sinks.
Jerry Martinson, I'm afraid you don't see an important distinction here. The devices you describe are not devices that move heat from a colder source to a hotter source without expending free energy. It is fine to spontaneously move heat away from a higher heat object, like a hot IC, towards a colder location. It is also perfectly fine for something that has the overall effect of moving heat from a hotter source to a colder source to, as a side effect, move some heat from a colder source to a hotter source. The issue is moving heat from a colder source to a warmer one without expending work (or, more to the point, without an increase in the entropy of the entire system considered as a whole).
An electrical diode violates no laws of any sort. When it is allowing current to flow, it is flowing from higher potential to lower potential, which is just what one expects. When it is not allowing it to flow, it is not spontaneously moving charge from a location of lower potential to higher potential without expenditure of work. This is all totally fine.
A "thermal diode" that permitted heat to flow from a colder to a warmer location would reduce the amount of entropy in the entire system. It would permit the creation of a perpetual motion machine -- you could run a heat engine between the hot side and the cold side, and allow the heat to "flow back" through the diode to the hot side, thus allowing the engine to operate forever. I could phrase this a dozen other ways, and it would all come down to the same thing -- you can't violate the second law of thermodynamics. The second law is, ultimately, a consequence of the conservation of energy, which is a consequence of the symmetry with respect to time of the laws of physics. It is not something that is going to happen.
I could imagine a device that permits heat to flow from a hot source to a cold source more easily in one direction than the other (though it would still need to be able to flow both ways to some extent). This might be an interesting sort of laboratory trick, though I'm not entirely sure what the applications would be -- if you want to move heat from a hotter object to a colder one you do not need a "diode" and if you want to move it from a cold object to a hot object you still need a heat pump. I cannot believe in a device that allows heat to flow from a cold object to a warm object -- that violates physics.
Back to err again. I figured there must be more to this story. The authors have excellent credentials and are working at a top school.
Reading what was given (don't have access to Science magazine) changed my mind. A little. First, I wouldn't call this rectification. The summary says a configuration was built that allowed heat to flow more readily in one direction than the other. Heat flow was not blocked in either direction.
It is not clear why two nanotubes of different material are involved. Were these two separate test devices? Or were the tubes both part of one device? And a platinum mass was deposited unevenly along the tube. I assume there was a gap to prevent the platinum from conducting any heat itself.
And how many devices were tested? If only one then error in the measuring instruments might well be suspected.
Still, it seems as if it does work. And thermodynamic laws don't seem applicable since power is being briefly applied to a heater at one end and the resulting heat then flows to the colder end of the device.
I still think that "Thermal Diode" or "Thermal Rectifier" is a pretty good term to me for this contraption, but maybe the perspective on appropriate semantics depends on your professional background. I think the researchers are trying to convey the importance of the potential applications of their work by using terms suggestive of how the device could be used as a component. I think your objections to this are that it is not an "ideal rectifier" so the term rectifier/diode would be misleading implying undue theoretical significance of the work concerning the 2nd law. However, to someone who is an Electronics Engineer like me who usually thinks in terms of circuit theory and components, I think that calling this contraption and how it behaves a "thermal rectifier" or a "thermal diode" is the best circuit component analogy to how this contraption behaves. When building or analyzing a complicated thermal system, it is useful to borrow the powerful analytical circuit-theory and system concepts that are very well-developed in the electronics realm and translate them using the following duals:
1. Voltage Temperature
2. Current Heat Flow
3. Capacitance Thermal Mass
4. Resistance Thermal Resistance
5. Voltage Source External Heat Source
6. controlled source Heat Pump, etc....
7. Inductance No good thermal analogy
8. Diode Thermosyphon, and this new solid-state device
To me, an electronic dual of "moving heat from cold to hot" would be a voltage source or battery, not a diode.
To my knowledge, most newly minted Electrical Engineer graduates are taught some quick "thermodynamics for EE's" that essential say use circuit theory above duals 1,2,3,4,5,6 to model/understand thermal systems. I have personally found this way of thinking to be extremely useful when modeling complex systems because you can use SPICE/Matlab/PDL to model electronics and thermals at the same time. Just like electronic schematics, a common pitfall is to not include all the parasitic components.
To be clear, practical electrical diodes are also not "ideal diodes" either and have a thermal noise voltage associated with temperature, Boltzmann's, etc.. via the Shockley equation. If it didn't, "electrical diodes" would also violate the 2nd law because thermal electrons would be rectified and able to produce useful work. Real diodes also don't fully block reverse current either, some leaks through. In some cases, circuits depend on this behavior. A "perfect" rectifier, whether it be electrical or thermal, would violate the 2nd law of thermodynamics.
My point was that thermal diode contraptions (even completely passive ones) already exist and are used in industrial applications. The most popular of which are often called "thermosyphons". These effectively behave in a thermal circuit just like a thermal diode and aren't much more complex to make than heat pipes and require no moving parts (other than moving vapor). There are some other applications of these sorts of contraptions that are very useful but I am limited in what I can say about them at this time. The existing ones are not solid-state and have similar limitations to what pre-transistor electronics. You could google "thermosyphon" and find many examples of this. What is novel in this article is that this is a "solid-state" thermal diode.
On an unrelated note, just like in the mechanical realm where there are proposed perpetual motion machines and Maxwell's demons, people have contemplated using electrical diodes to make usable energy from pure heat (see U.S. patent 3,890,161). As far as whether someone someday will be able come up with a contraption that violates the 2nd law as we currently interpret it, my guess is that, analogous to Shannon's theorem, we'll probably see some contraptions and devices that will narrow our definition of what "noise", "random", "entropy", and "heat" truly mean as our statistical understanding of these things is still rather primitive but the law will essentially hold.
Very quick note. I'm going to look up thermalsyphons.
But Jerry seems to be talking about using energy or convection to remove heat and/or about heat sinks. And Perry seems to be talking about what heat does when left alone - it moves from hotter to colder material.
What is a 'passive' device to Jerry?
I don't know about Jerry, but to an EE a passive device is any device that does not contribute power to the circuit and thus requires no power source for its operation. Transistors and all other amplifiers do require a source of power.
If you search online you will find some articles that incorrectly classify a diode as active; however, a diode is simply a resistor with a variable (exponential) resistance. That said, I suspect many EE curricula probably include diodes in a course on active circuit analysis rather than a course on passive circuit analysis because the iterative methods used for solving the equations for diode circuits resemble the methods used for transistors.
In the same way as an electronic diode does not force current to flow in the 'wrong' direction, a thermal diode would not force the transfer of heat from a cold source to a hot one. It would purely act as a material which would allow the transfer when it was in the 'right' direction, and insulate tranfer when the natural flow was in the 'wrong' direction.
Personally, i think this would be ideal in situations when you want passive heating or cooling when it is availible, but want to maintain a physical barrier to the outside.