reason why liquid nitrogen kils proformance

[imposter]

Part 1, Part 2 with explanations will be posted at the end of this.

Hey all, Yuriman and me were talking on AIM the other day and got into a large fight (as usual) whether LN2 (liquid nitrogen 2, liquid nitrogen is diatomic always) would increase or decrease performance in cooling. Yuriman believed that Liquid nitrogen makes wires superconductors. I kept telling him that's impossible and here is why I am correct and he is wrong.

I basically set up an experiment with my Science teacher in school. Since she was going to show liquid nitrogen to the class I asked her if I may use some of it to run an experiment while showing the class that liquid nitrogen makes metal more resistive making bad conductors. She agreed.

So I set up my experiments, but I can't show you pictures right now because I don't have a digital camera. But, I do have this one paint picture/ diagram that may help you understand better.
This is basically is a diagram in what I did.

a= battery
b= containor fully of nitrogen
c= fan
d voltmeter
e= where i spliced.

I have got a 6 volt battery (one the big ones ;-)). I attached the alligator clips to it (part A), then connected a 12 volt care fan to the other side and stripped the wire in between so I could get my voltage measurements. With my voltage meter, I got my first reading which was 6.4 volts.

After getting my first measurements I put the middle of the wire into part B which held the liquid N2 and then waited a while, that way it could cool down. FYI this is -210C (Since it was in a cool container)

I then got my next reading of 6.2 volts. Clearly showing .2 losses in conductivity, even more so when I figured out that it want to be in the liquid nitrogen all the way, seeing we were running out. Knowing how elements behave

Reason for less performance, has to do with the element in this case copper and the electro activity of that element. It has to do with the electrons of the elements and as you slow down the electrons causes them to move slower causing decreasing performance. More will be added to soon…

most people think that liquid nitrogen makes metals super conductive. that is backwards the Ln2 makes magnets and magnets only super conductivity. magnets That is why when you go and get an M.R.I. done, that magnet is about $30,000 if not moreand is stored in liquid nitrogen. The colder a magnet gets the more conductive it becomes.

 

[El<(')>Maxi]

An element being in a superconductive state does not mean it will flow more or a higher current. What it means is that DC current will flow with zero resistance, basically in an infinite loop in a closed system. CU is also not a superconductor according to this.



Still thats a very interesting experiment you made

 

[Flip-Mode]

My dad explained this to me once, he said that at super cool temps the electrons slow down thus reducing the resistance of the material.
I'll ask my GF tonight, shes engineer student she should know...

 

[>HyperlogiK<]

If you mean CPU cooling people do use LN2 to cool processors, but then CPUs aren't made of wires:
http://www.tomshardware.com/2003/12/30/5_ghz_project/
I think that the current super pi record is held by a japanese team who had an intel 65nm chip at 7ghz.

I suck at physics but I'm fairly sure of most of the following:
LN2 definitely doesn't only make magnets superconductive, as Maxi's table shows (although MRI scanners do use magnets made from superconductors). LN2 won't make most potential superconductors (the regular metal ones) superconductive, because it isn't cold enough, they need to be cooled below 10K, and some of them to within a fraction of absolute zero. Cooling a metal with LN2 also definitely doesn't increase resistance either, in metals resistance falls with decreasing temperature (to a certain point), because the low temp reduces the vibration of the metal atoms. However I do think low temp reduces the number of free electrons, but that has nothing to do with resistance.

Would somebody with a grasp of quantum mechanics please post an accurate answer

 

[Prandtl]

You should have used a resistor (with known value) for your test and measured the current in the loop, not the voltage. Also, one thing to consider is that most battery will suffer from a voltage drop depending on the amp you draw. Any ideas on the battery spec? A good test setup will have require a voltmeter for the battery and an ampmeter.

I am pretty convinced that the resistivity of copper drop with temperature.
read me

 

[soddemfx]

The number of free electrons has everything to do with resistance in a semiconductor!

In a metal conductivity increases with decreased temperature. With all due respect whilst you had the best intentions (and credit for trying ), your experiment has errors.

A better approach would be to wrap 10m of very thin coated copper wire around something small like a match box. Then measure its resistance and place it in series with a resistor close to the same value. Now attatch you battery and measure the voltage in between the coil and the resistor. Place your coil in the nitrogen dewar and measure the voltage again, it will be different with more voltage being dropped across the resistor. With this you can calculate how much the resistance of the copper has decreased as its just a potential divider

In a metal resistance depends on the net electron velocity, which is determined by how easily the electrons can move through the metal. Under high temperatures there is much thermal excitation energy introduced into the lattice which causes the atoms to "vibrate" and move from their normal positions. Simply put the electrons moving due to the presence of our electric field collide with these vibrating atoms and slow down. This impedes their progress through the lattice, decreasing the net electron mobility, therefore increasing resistance.

When the metal is cool, the atoms do not vibrate so much, so less collisions occur, so the net electron mobility increases, therefore resistance decreases.

In a semiconductor decreased temperature decreases electron population in the conductance band, dropping to the valence band and decreases electon mobility, therefore drift velocity decreases, therefore forward drain current decreases. For example under low temperatures the number of free change carriers in the channel of a FET per unit time is reduced so the resistance essentially increases.

Another problem is the band bending between the metalic interconnects and silicon caused by the low temperatures.

The only ways i think lower temperature benifit us are:

Reduced propagation delay within metalic interconnects.
Decreases in gate oxide leakage.

I dont know about superconductivity as all semiconductor work ive done was at room temperature -but i do think that the term "superconductivity" is used far too much for the wrong reasons.

Tom

 

[imposter]

Im sorry that i don’t have all my hw before this post was up I was to tired today to finish it I should have the rest of my Info posted either late today or sat I have a test coming up on Monday that I got to for study that. My argument will make a lot more sense when I post up the information. I also am going to have to reconfirm this info before I post it as well which is going to take a while. I should have the draft of my info by tonight if not 2moro in class. I will review it with my teacher and make sure it’s accurate. Before I post.
And Im not sure but if I remember correctly U Can use a formula to determine Resistance base on voltage. AND I did use a volt meter/amp/ resistance meter. And for your info The Amperage is not something u can really test on a line. You can only test how many amps are being used. That’s why I had the fan in there. I would have done strait Resistance test but I had no clue how to work the thing. So I gave up on it. And again as a test the Line before anything was attached the battery gave off 6.42 Volts. And I got about 6.4 volts on the other end of the copper wire. Sorry I forgot to mention that. In the process of revising this message.

I dont know about superconductivity as all semiconductor work ive done was at room temperature -but i do think that the term "superconductivity" is used far too much for the wrong reasons.

Tom

i agree what you said in the last paragraph.
also in the process of asnwer to your comments please allow me some time Im very busy atm.

 

[Prandtl]

And for your info The Amperage is not something u can really test on a line. You can only test how many amps are being used. That’s why I had the fan in there.

Use a straight resistor, not a fan. A 50 ohm or less if at all possible. Review Kirckhoff (spelling) law. You can measure the current (amperage) in a loop, you only need to place the ampmeter in serie (sp) with the other element..
ie:
+side of battery ---> LN2 tank ----> + side of amp ----> -side of amp -----> resistor ----> LN2 tank -----> - side of battery
Sorry if I am stating the obvious, which I probably am, but I cant help it

 

[Yuriman]

The word "fight" is a bit too strong.

 

[GigaForce310]

The only superconductors I know of that are not specific alloys won't become super conductors at -195C. There are certain alloys that become superconductors at over 150 K or about - 123C. (After researching a little bit I now recall that the numbers I stated are incorrect. The highest temperature superconductor known happens below 133K. HgBa2Ca2Cu3O_8)

The reason LN2 gets better results is that the field effect of enhancement mode Metal Oxcide Semiconductor Field Effect Transistors (eMOSFETs) is enhanced by reduced state switch latency. In laymans terms is that -195 C when it doesn't cause other problems, lowers the time each of the hundreds of millions of transistors takes to transmit data.

There is a point where the temperature gets so low that the process slows down. This happens somewhere under -220 C. LN2 turns out to get near the coldest a processor can take before the cold starts to slow electron movement nearing the drift velocity. At room temperature, electrons move a something like 1% of the speed of light. What matters is how much they drift on average. If the electons slow way down then eventually the drift velocity will suffer and the performance of the processor will slow. Liquid helium doesn't have enough latent heat and liquid hydrogen is very dangerous to work with. In theory they both are able to slow a processor down, but in practice it's not feasible.

With the resistances, I don't know what they do at cold temperatures except superconductors lose resistance. The resistances in electric circuits don't cause any more problems than more friction slows a car. Add a little more push (voltage in this case) and that problem is accounted for. That could be the reason that slightly higher voltages are safe with very cold cooling.

 

[soddemfx]

The reason LN2 gets better results is that the field effect of enhancement mode Metal Oxcide Semiconductor Field Effect Transistors (eMOSFETs) is enhanced by reduced state switch latency. In laymans terms is that -195 C when it doesn't cause other problems, lowers the time each of the hundreds of millions of transistors takes to transmit data.

?

To switch on a MOS transistor a voltage Vg is applied to the gate such that Vg = threshold voltage Vt. This causes extensive band bending which in turn causes the onset of strong inversion and the creation of a conducting channel.

This is the structure of a MOSFET:

This is a cross section of the gate oxide reigon:

Vt is a product of Vs (voltage across depletion region) and the voltage across the oxide, Vo. Vg = Vo + Vs.

From the band diagram when the device is at the onset of inversion:

If my word diagram wasnt so bad you could see that -qVs is half the flat band work function (PHI_fb).

So Vs = 2PHI_fb/q

It can be proved that:

Ev - Ef = kT ln (Na/Nv)

Since:

Ei - Ef = PHI_fb

PHI_fb = (Eg / 2) - kT ln (Na/Nv)

So Vs = 2((Eg / 2) - kT ln (Na/Nv))/q

It by use of the poisson equation it can be proved that the width of the depletion region:

dp = ((2EE₀Vs)/q Na)^1/2

As Qb = Cox Vo

where Qb = q Na dp

Qb = q Na ((2EE₀Vs)/q Na)^1/2

So Vo:

Vo = (q Na ((2EE₀Vs)/q Na)^1/2) / Cox

sub in our oxide capacitence gives:

Vo = ((q Na ((2EE₀Vs)/q Na)^1/2) Dox) / EE_ox

Vt = Vs + Vo

Vt = (2((Eg / 2) - kT ln (Na/Nv))/q) + (((q Na ((2EE₀Vs)/q Na)^1/2) Dox) / EE_ox)

Using:

Na = 3 x 10²⁰
Dox = 100nm
E_si = 11.9
E_ox = 3.8
Eg = 1.11ev
Nv = 1.08 x 10²⁵

Assume T = 300k

Vt = 0.793v

Assume T = 78K (liquid nitrogen)

Vt = 1.22v

That is seriously not good. No wonder extreme cooling requires more voltage...

Briefly electron mobility from "Solid State Electronic Devices, 2000":

"On the other hand, scattering from crystal defects such as ionised impurities becomes the dominant mechanism at low temperatures. Since the atoms of the cooler lattice are less agitated, lattice scattering is less important; however, the thermal motion of the carriers is also slower. Since a slowly moving carrier is likely to be scattered more strongly by an interaction with a charged ion than is a carrier with greater momentum, impurity scattering events cause a decrease in mobility with decreasing temperature...

...the approximate temperature dependencies are......T^3/2 for impurity scattering"

Electron mobility is a large factor in forward drain current, and that is a large temperature dependance. Increasing voltage would probably compensate low temperatures up to some point.

Tom

 

[matttheniceguy]

soddemfx, I think you need to post more often. Good (although a bit hard to follow compleatly) info.