• Quick note - the problem with Youtube videos not embedding on the forum appears to have been fixed, thanks to ZiprHead. If you do still see problems let me know.

Would a superconducting material...

It doesn't seem at all possible. Heat is the kinetic energy of the molecules in the material. This means heat can't travel across a material without making the molecules vibrate faster, but if the molecules have increased their vibration, they must have absorbed some heat!

Another way to look at it is by the first law of thermodynamics: Heat travels from hot to cold. How can one end of a heat pipe be hotter than the other end if it doesn't absorb some heat? A material can't be a perfect thermal contuctor if it absorbs heat!

Anyway, that's how it seems to me. I've heard about some pretty strange carbon-fiber materials that conduct heat in one direction (along the fiber), but block heat in another (across the fiber). Maybe something has changed in the last few years since I dealt with such things.
 
falzer said:
Would you mind saying what your source for that information is? I'm curious because the only place I've seen that was in a Niven novel, and Wikipedia's article on superconductivity suggests that superconductors do not have infinite thermal conductivity.
Click to expand...

erm... blush... that may be where I got it from too... lemme check...
 
I should add that conductivity/resistance isn't the main source of heat in a computer, switching losses are.

Power dissipated across a transistor is determined by the voltage drop across the transistor and the amount of current through the transistor by the equation:

P = I x E

This is just the old P=IE, or PIE equation that is taught in a basic electronics class. So if we know the voltage and current through a transistor at any given moment, we know the power loss created by the transistor.

When a transistor in a computer circuit is off, it has full voltage across it but no current flowing through it, so the power dissipation is essentially zero. I say essentially because there is going to be some tiny leakage current but the resultant power loss is negligible compared to the switching loss. The power dissipation is P = I_min x V_max = 0, since I_off is pretty much zero.

Similarly, when a transistor is fully 'on', it has maximum current I_max flowing through it, but a very low voltage drop. Again, the power dissipation P=IE is low because the voltage E is low. In this state, there is a little more power loss because it is hard to get the voltage drop exactly equal to zero, harder than making the current flow zero in the 'off' condition. The power dissipation is now P = I_max * V_min = zero (pretty much).

Now suppose the transistor switches from 'on' to 'off'. The transistor needs some time to make the transition, so halfway through the process, the voltage is about half its maximum value and the current is at about half its maximum value. The power dissipation is now

P = I_max/2 * V_max/2
= (I_max * V_max)/2

Since I_max and V_max are pretty significant numbers, the power dissipation is significant during the switching transition. The longer the transistor takes to switch, the more time it spends in this switching region and the more heat it develops. Also, the higher the switching frequency (on/off cycles per second) the more heat is generated per unit time. This explains why a computer CPU needs a big fan or even liquid cooling when it is overclocked (run at a higher than design clock speed).

The amount of power dissipation can be minimized by reducing the switching time. Of course, this only tempts electrical engineers to increase the clock speed of the CPU and you're back where you started with heat generation. The most productive methods to reduce power generation are to designe the CPU to draw less current when the transistors are switched on. Also, a lower power supply voltage will make cause the switching losses to be lower. In the old days, computers ran off of 5V but now energy-efficient CPUs used in laptops use a 3.3V standard (I think). I'm pretty sure memory also uses a lower voltage but I'm too lazy to look this up for myself.

Anyway, there really isn't a way to cut switching losses to zero unless someone figures out a way to make transistors draw zero current when they are turned on. In theory, there are ways to do this, or at least, come very close.
 
rocketdodger said:
No. Virtually all of the heat generated is on the chips, which won't be replaced by superconductors anytime soon. Furthermore all the resistors can't be superconductors by definition, so those will always produce heat.
Click to expand...

Aren't vacuum tubes transistors? And don't they work by charged plates, rather than semiconductors? I know it might take a bit of work to build that in as a printed circuit component, but I wouldn't say it can't be done...
 
Transistors are by definition semiconductor devices. Tubes operate by a different principle, involving emission of electrons from a heated, charged metal surface toward another surface. Are there any superconductors that work at high (not enough to cause severe skin and flesh burns) temperatures?
 
Terry said:
erm... blush... that may be where I got it from too... lemme check...
Click to expand...
The specific heat of a superconductor is not zero, but is much smaller than what a normal metal would have.

At low temperatures, the specific heat of a normal metal goes as c = A·T + B·T^3 (the linear term being conduction by electrons and the cubic term by phonons, i.e., lattice vibrations). If we drop the temperature below Tc (the critical temperature for the superconducting phase change), c increases and then falls much more rapidly than what the previous formula indicates (exponentially). If we now apply a strong magnetic field (Bc), the superconductivity is destroyed
[*] and we can measure the specific heat of the normal metal at that temperature. It is much higher.

The precise form of the superconducting heat capacity can, on ocassion, be explained by an energy gap. This is consistent with the BCS theory of superconductivity. Keep in mind, however, that there are many kinds of superconductors and this theory does not work for all of them.

If you want some rough explanation of why superconductors have small specific heats, you can think of them as being more ordered (lower entropy).

Edited to add:

Ouch! I now realise you were talking about thermal conductivity, not specific heat. Superconductors, contrary to what one would think, are poor thermal conductors and don't exhibit the Peltier effect. This indicates that the conducting electrons don't carry entropy. As before, we can measure thermal conductivity at a given temperature for the normal and superconducting states by application of a magnetic field and we see that the normal metal is a better heat conductor. This poor conductivity makes them useful for making thermal switches.

All this seems contrary to intuition: we have learned that good electrical conductors are also good conductors of heat (after all, the electrons carry both current and heat). But this doesn't work with superconductors, because in them the electrons are not independent, but coupled.

________________

[*] At very low temperatures and zero magnetic field, some metals are superconductors. If we apply a sufficiently strong magnetic field they become normal conductors again. At each temperature there is a critical magnetic field, Bc. At the critical temperature for the superconducting phase transition this critical field is zero. With some materials (Type I superconductors), the change is very sudden. Type II superconductors experiment a transition state (vortex state) and have two critical fields. This is quite important for the practical applications of superconductors (magnetic resonance machines in hospitals use superconductors to generate their magnetic fields, if those superconductors had a small Bc they would be worthless).
 
Last edited:
bjb said:
It doesn't seem at all possible. Heat is the kinetic energy of the molecules in the material. This means heat can't travel across a material without making the molecules vibrate faster, but if the molecules have increased their vibration, they must have absorbed some heat!
Click to expand...

Not quite. Heat is basically randomized energy in the system. In a solid, that often involves the atomic nuclei vibrating rapidly, but it ALSO can involve the electrons becoming excited as well. In insulators, the electrons need to overcome an energy gap, and so do not play a significant role in either the heat capacity or the thermal conductivity, but in metals, there is no such gap, and so the electrons are not only very important to the heat capacity, but they actually dominate the thermal conductivity. And that's because excited electrons can move around. That's why good electrical conductors are generally good thermal conductors as well.

So naively, it's actually not unreasonable to think that a superconductor would make a good thermal conductor as well. But it doesn't work out that way. Superconductivity is driven by Bose condensation of electrons, where the electrons enter a particular quantum state where they are paired together. When paired up, these electrons have an energy gap to any excitations, and so do not scatter off the atomic lattice as they do in a metal. This means they can conduct electricity without any resistance, but for the same reason, they also don't conduct heat well. From a thermal conductivity perspective, they act more like the electrons in an insulator.
 
high temperature superconductors

Research is underway to build high speed transistors using gallium nitride. Some devices (transistors and, I think, a simple op-amp) have been built and operated at 350C. So rather than superconductors for a cooling path, you could hopefully remove the need for cooling altogether. I know a little about this because of a proposal I put together a couple years ago to build rf power amps on gallium nitride...proposal was unfunded and I moved on to other stuff.

BTW: don't hold your breath on these things...gallium nitride crystals are very hard to grow of sufficient quality...but there's a couple companies working on it (company called Cree, and one called Nitronix, I think).
 
So theoretically at least, a vacuum tube (aka valve) employing superconductors might be possible. Whether it'd have useful propeties... *shrug* that's way beyond me.

[edit] Oops, I think I misinterpreted StuBob's post caption and spoke too soon. From what I see by googling, gallium nitride's a semiconductor, not a superconductor. Different critters.
 
Last edited:
semi not super

ya got me Mef...I was thinking semiconductors and typing superconductors. Homer Simpson lives in me! :)
 
Ah well, I wouldn't be surprised if some previously overlooked niche of physics makes high-temp (as in hot) superconductors possible eventually.

[edit] Isn't gallium nitride a LED material? *googles* Yes, blue, and with other dopants (like indium) they're getting a pretty wide range of wavelengths.
 
Last edited:
Meffy said:
Ah well, I wouldn't be surprised if some previously overlooked niche of physics makes high-temp (as in hot) superconductors possible eventually.

[edit] Isn't gallium nitride a LED material? *googles* Yes, blue, and with other dopants (like indium) they're getting a pretty wide range of wavelengths.
Click to expand...

There's no telling what THEY will come up with :)

I would so love to see what is in use 100 years from now. When you look at the last 100 years, and the huge advances that have been made, I think most of us (me at least) would have a hard time imagining what folks will have in their hands, on their desk, or are driving around 100 years from now (or what has replace hand held/desk top/transportation devices).

Compare 1906 tech with todays. Then try to picture 2106!:rolleyes:

Not an attempt to derail, it's just something that crosses my mind every now and then. And I wish there were true future tellers, with video cameras, to show us ;)
 
Certain psychics claim to have pens that write messages from the future, but no tellers.
 
Interesting answers, as always!

So, basically, a superconductor would not really have any impact on PCs at all, as the only things you could really replace are the pathways between components (like the bus) and those aren't really producing much heat anyway... The reason I thought about this, was the ever-increasing power consumption of video cards and CPUs and their need for more powerful cooling techniques. Heh, if I remember correctly, I don't think my old 386 had any fans at all!

Still, a room-temp superconductor would be good news for the power companies, I suppose, what with their endless miles of heat-leaking power lines.
 
Raskolnikov said:
So, basically, a superconductor would not really have any impact on PCs at all, as the only things you could really replace are the pathways between components (like the bus) and those aren't really producing much heat anyway...
Click to expand...

Oh, I don't think I'd go that far. One of the dominant factors in computer speed is the pathways between components, which in turn is dominated by the resistance and capacitance of the wires. A signal will propagate much more cleanly and much more quickly in a superconducting wire than in a normal metal wire, which in turn means that you run a faster and sharper signal.

I'd not be surprised to get a 10x speed improvement out of a such a system.

Resistance covers a lot of stuff besides heat.
 
drkitten said:
Oh, I don't think I'd go that far. One of the dominant factors in computer speed is the pathways between components, which in turn is dominated by the resistance and capacitance of the wires. A signal will propagate much more cleanly and much more quickly in a superconducting wire than in a normal metal wire, which in turn means that you run a faster and sharper signal.

I'd not be surprised to get a 10x speed improvement out of a such a system.

Resistance covers a lot of stuff besides heat.
Click to expand...

Yep. This was also thought to be an advantage of optical systems, even if you still have to use electric ICs. Of course, the expense makes it nto worht the effort.

Also, assuming true room-temp superconductors, all your power cables and wires would benefit. Better transmission of power, with less heat and loss there. Not to mention the general power grid, but that's a bit outside the computer discussion.
 
You must log in or register to reply here.

Back
Top Bottom