Re: How does a thermocouple work in terms of electron flow at junctions

I sense from your message that you are having some difficulty with
the subject of thermoelectric effects. You are not alone; these are
not trivial subjects and the physics behind them is not simple.

I must start off with a disclaimer. There's no way I can cover the
subject in great depth via email. I can give you some pointers and
a brief introduction, but there's no way I can write a book here,
and the subject will require that you read and understand at least
several chapters from texts.  That said, I was somewhat surprised to
find that all of the solid state physics books I consulted (5 of them)
were rather sketchy in their treatment of thermoelectric effects, and most 
of what was there concerned the suhject in semiconductors rather than 
metals.In any case, let's start off with a few references and other places
to go, just in case my short exposition leaves you unsatisfied. There
are a number of web references (usually by companies that make thermo-
electric devices like Peltier refrigerator modules or thermocouples),
one book, and the International Thermoelectric Society web address.
The latter may serve as a gateway to find real experts in this field.

   On-line documents of some use:

     http://www.tellurex.com/resource/txfaq.htm
     http://www.ferrotec-america.com/3refframe.htm
     http://schottky.ucsd.edu/~felix/peltier.html
     http://www.zts.com

  The address of the ITS:

     http://www.its.org

  Book:

     Kittel, "Intro. to Solid State Physics", 6th or 7th ed, Ch 9 and
       a brief section on semiconductor thermoelectric effects in Ch 8.


Let's start with a couple of experimental facts. Each metal has a certain
density of conduction (essentially free to move) electrons in it, and each
metal holds onto its electrons with a different amount of "eagerness". We
can quantify the latter by firing photons at the metal and seeing that at
a certain photon energy, electrons start being knocked loose (this is the
photoelectric effect).  The amount of energy required to knock and electron
loose is called the workfunction of the metal.  Let's say, for example, 
that we have a metal with a workfunction of 4.1eV and another with a 
workfunction of 4.7eV.  If I pull an electron out of the 4.1eV material and 
drop it into the 4.7eV material, I release 0.6eV.  Of course, when I do 
this, I also transfer an electric charge, making it harder to transfer the 
next electron. Now imagine butting the two metals together. Because energy 
is released by going from the 4.1eV workfunction metal to the 4.7eV 
workfunction metal, electrons will spontaneously transfer that way...
at first.  As they do so, though, they separate charge and cause an
electric field (in the direction to oppose further electrons from crossing
the boundary) to build up.  Now, just to make life difficult, Mom Nature 
also imposes a third effect. Since the density of electrons in the two 
materials is usually not the same, there is also a diffusion current driven 
by the concentration gradient of electrons.  So we have three things going 
on: diffusion current, drift current (from  the induced electric field) and 
the initial intrinsic workfunction differences.  

These three effects eventually sort themselves out and a state of 
equilibrium is achieved with a certain amount of charge separation between 
the metals and no net current flowing (electrons still easily and readily 
cross from one metal to the other, but in equal numbers in each direction). 
However, there is an electrical potential diffenence in there, and this is 
a very real effect. (For example, it causes corrosion to happen if, say, 
iron and copper pipes are put directly in contact with an electrolyte 
(water) flowing through them). If you put a voltmeter across the two 
metals, however, you will not read anything because everything is in 
equilibrium and the electrical potential is balanced out by other forces 
(e.g. the concentration-driven diffusion).

But...it turns out that the equilibrium is highly temperature dependent. 
That is, the balances shift as we raise the temperature.  Certainly 
electrons are more energetic at higher energies, and this is seen as 
changing values of the workfunctions (get hot enough and electrons will 
boil out of a material all by themselves without any photons being 
necessary. Of course, it may melt, too, in which case we're sort of out of 
the domain of solid state physics...:-). The math behind all this is 
somewhat obnoxious but to get a really solid understanding of all this 
you're going to have to come to grips with it. You should get familiar with 
the concepts of Fermi levels,  the density of states in metals, free-
electron gas energy statistics, and the drift and diffusion currents in 
materials, at a minimum. 


Now, let's make a loop of wire out of two metals, say Pt and Cu.  Clearly,
the Pt and Cu meet at two points. Lets call these the Hot and Cold 
junctions, because we're going to heat one of them and cool the other.  Two 
items  should be apparent. One, since hotter electrons move faster than 
cooler ones, the electrons will tend to be, on the average, diffusing away 
from the Hot junction more than toward it. This tends to cause a diffusion 
current away from the Hot junction and toward the cold one. We will also 
note that the internal junction potential barriers are different at the two 
temperatures.  The net result of these phenomena is that a small potential
difference exists between the junctions and some current will flow. We can
read the potential difference and calibrate the amount of it vs. the 
temperature. Hence, the thermocouple.

There are other fascinating thermoelectric effects as well, especially the 
Peltier effect, wherein running a current through a loop of two dissimilar
materials causes one junction to cool down and the other to heat up. Some
of the web references have explanations of how that works.

Hope this has been of some help!