john harper:

> >>We used to call it "contact potential."
> >
> >You are right that there exists a bit of contact potential. But it's not
> >very much in this case.
> >
> >Contact potential is on the order of 0.1-0.2 microamps. That means with
> >a 10 megohm grid resistor you get 1 to 2 volts of negative grid bias.
> >This is a fine amount of bias for a 12AX7 or other hi-mu tube.

> Can someone explain the connection between "contact potential" and the
> mu of the triode used?

There seems to be some confusion here. "Contact potential" is a voltage, not a current (hence the name), and it refers to the effective grid voltage resulting from the difference in metals used for the grid and the cathode. It is in the range +0-0.5V (volts, not amps).

A more signifcant reason for an effective positive grid bias in the absence of any actual bias is the velocity of emission of the electrons. At low currents this is around 0.7V. There is a "virtual cathode" just ahead of the real cathode which has this potential (i.e. -0.7V), making the unbiassed grid look positive, made up of the electrons which are emitted but which do not flow to the anode. (You can clearly see this in the curves of high-mu triodes such as 12AX7, where the characteristic positive-grid kink is present at 0V grid bias).

If you leave the grid truly floating, it will attract electrons until its potential is just enough to counter these effects, i.e. it will float at around -0.5 - -1V.

Under normal operating conditions things are different. The grid attracts positively charged ions resulting from electrons banging into residual gas, resulting in a negative current flow (i.e. one tending to make the grid more positive). This current is a function of two things, the plate current and the amount of residual gas. In fact, measuring it is a way to determine the level of gas in a tube. If the tube's mu has any effect at all, it is very minor.

john harper:

I'm afraid this is a pretty jumbled version of what is going on. It gets close to the right place but the way it gets there is not right.

> Whoa here, boys. I know the RDH has the unfortunate sentence about
> ""contact potential" being an electrochemical phenomenon". As far as I can
> tell, this is due to quite a bit of jumbled thinking.
> There is no electrochemistry going on.

True there is no electrochemistry, i.e. no there is no primary cell generating current or similar. But it is nevertheless correctly referred to as an electrochemical effect. Any two conducting materials have a contact potential between them. This is the principle behind all chemical batteries. If there is no electrolyte (as in a valve) then there is no current, but the potential difference is still there. It is due to the different energy levels of the valence electrons, and since chemistry is essentially about valence electrons it is electrochemical. (Don't go looking for pretty colours in books about quantum chromodynamics, either).

> Now INSIDE the tube, there is another potential developed, much larger and
> from totally different mechanisms. It has nothing to do with
> electro-chemistry or contact between dissimilar metals.
> It's due to the Bose-Einstein statistics of the electrons emitted from the
> filament or cathode (No not OUR Bose). A certain percentage of the
> electrons have more energy than the average and they make it thru the
> electron cloud and land on the grid. Electrons on the grid make it more
> negative. There is a very small number of electrons that make this trip,
> and they have a very limited voltage, so there are very definite limits on
> both the voltage and the current developed. It is by no means a definite
> potential with limitless current behind it. It is neither due to "contact"
> nor a realy definite "potential". But since it doesnt matter much, the few
> people that still discuss it these days call it "contact potential",
> sometimes remembering to put it in quotes but often forgetting why.

Well now. True that there is an energy distribution among emitted electrons, although I don't believe this has anything to do with Bose-Einstein statistics (although quantum mechanics is necessary to derive the distribution). But they don't "land on" the grid. A tiny proportion of the emitted electrons (in normal use) has sufficient energy to penetrate the potential barrier set up by the space charge ahead of the cathode. These are the ones that reach the anode. If you reduce the anode potential to zero, a small current still flows because the virtual cathode is actually negative relative to the physical cathode, and hence the anode still appears positive. If the grid is ALSO held at zero volts, then a small positive current will flow in the grid, since it too is effectively positive wrt the virtual cathode.

To reduce plate (and grid) current to zero is in principle impossible. However if you consider that anything under a microamp or so is as good as zero, then you need to reduce the plate potential to a small negative amount - say -1V or so.

If the grid is left disconnected, then electrons will flow to it until it reaches approximately the same voltage as the virtual cathode (plus the contact potential due to the differing materials).

If the grid is held significantly negative wrt the physical and virtual cathodes, which is the normal case, then positive ions resulting from collisions with residual gas are attracted to it, resulting in a small but measurable negative current. This is why a grid resistor is essential - without it, the grid would gradually become more positive until it reached the balance potential of ~-1V wrt the cathode.

Probably nobody except George and myself is interested in this anyway...

Ken Gilbert wrote:

> > Probably nobody except George and myself is interested in this
> anyway...

> Not quite true, John... I am following with interest.

> A somewhat related question I have:

> I've generally read to never leave elements in a tube floating... what
> is the potential harm in this practice? I don't see any current paths,
> and if there's no current then there's no power dissapation, so what's
> the deal?

The problem is that it isn't really "disconnected" since it is in the path of the electron stream. It will therefore attract electrons and ions until it reaches a balancing potential where it is attracting them in equal quantities. In the process it will affect the electron flow exactly as if it was tied to the same fixed voltage. This is probably not whatever you wanted. Offhand I see no reason why this shouldn't be the same as the voltage for the control grid, i.e. around -0.5 - -1V. A screen at this potential will certainly not give what you were expecting!

ken gilbert:

Thanks for the reply, John. Indeed, I found that when the screens were floated, the current through the tube ceased, as would be expected with a screen voltage of around 0. I suppose I could measure it, just for the hell of it.

It came about due to the 4PDT switch I used for triode/ultralinear switching for two channels. The center position is N/C, which pinches off the tubes.

Again, thanks.

gary longrie:

> > So is is the same as the thermocouple effect? Are we talking about
> > nickel grid wires vs. barium oxide cathode? So do we have an evacuated
> > dry cell with electrons in a vacuum instead of ions in solution as
> > current carriers. Velly interesting. Can we make a 3-volt contact
> > potential by using lithium?
> >
> In effect that's it. I guess in principle yes, you could use Li to
> get a higher contact potential, although your tube probably wouldn't
> last very long!
> John

greetings guys,

Thanks for a very interesting thread. You made me go dig in the books and library for awhile. Here's what I find. This is fairly essential stuff if one is really going to get in there and understand tubes at the physical electronics level. John, you are working on models these days aren't you? This could have implications for modeling as well. I think that the rat and aga tube communities would be interested if enough clarity could be provided in describing these concepts. So please allow me to have a try at it. This is what I come up with. Naturally, it may benefit from further discussion.

An activated barium oxide cathode is operating at dynamic balance between a pair of chemical processes. Barium oxide by itself is not a particular useful electron emitter. During the activation process, enough barium oxide is reduced to provide a sufficient amount of pure barium, which is an excellent donor of electrons. So the trick is to maintain just enough pure barium in the cathode. Gas and cathode impurities will tend to react with barium and oxidize it. Other reducing gases and agents will react with barium oxide and reduce it back to pure barium. These oxidation and reduction processes work against each other in slow time. Gradually the oxidation process pulls ahead and the emissions rate slowly declines, over thousands of hours.

The work function of a metal is the energy, in electron volts, that must be given to an electron at the Fermi level to enable it to escape from the surface of the metal. The potential in volts is numerically equal to the work function energy in electron-volts. The assumed conditions are that the electron escapes with zero velocity and that there are no external potentials or fields. Work function has three different definitions.

The true work function of a metal is a characteristic of that metal at absolute zero degress kelvin. But itís kind of a misnomer. The true work function is a single data point valid only at a single temperature. It's the value given in the lookup tables. The true work function is used in the Richardson-Dushman thermionic emission equation.

The work function of a metal is slightly affected by temperature. At temperatures above absolute zero, some of the electrons occupy higher energy levels, and take less excitation energy to escape. Starting with the true work function calibration point at absolute zero, the temperature coefficient is something like 10exp-5 or 10exp-4 ev/deg K. When the work function takes temperature into account, the result is called the effective work function. The effective work function is preferred because of its correction for temperature dependence. See Kohl for lots more references on this.

Sometimes called the apparent work function, the Richardson work function is an older, averaging, graphical technique using the slope of the Richardson plot J/T**2 vs 1/T. It is assumed to be independent of temperature.

In a metal, the Fermi level is the maximum energy that an electron can have at zero deg K. This is to say that all the electrons have taken the lowest available energy levels. Then the electron with the highest energy level is at the Fermi level. The difference between the Fermi energy and escape energy is the work function.

Sometimes called the contact difference of potential or contact potential difference, this one is intuitive and obvious. From the definition of work function, we can see that the contact potential difference between two metals is numerically equal to the difference of the work functions of the two metals. If several wires made of different metals are connected in series, the potential difference at the two resulting ends is just the difference in work functions of the two end metals, taken in volts.
When a voltage difference is applied between two electrodes of a tube, the electric field intensity in the interelectrode space effectively results from the sum of the applied potential difference and the contact potential difference. One can readily see then that if a grid is biased a few volts from the cathode, the electric field between grid and cathode would greatly influenced by a contact potential of a volt or two.
A part of the aging, or drift, in tube characteristics has been attributed to a change in contact potential difference between grid and cathode as a result of gradual contamination or decontamination of the grid surface.

Since an oxide cathode is a semiconductor, there is a very real resistance of several ohms. As the cathode ages and the active barium becomes depleted, the impurity level rises and the cathode resistance increases.

The barium oxide reducing and oxidizing reactions take place at the interface between the oxide and the backing nickel sleeve. The reduction reaction products are free barium and an impurity oxide. The free barium diffuses back to the oxide emitting surface where it serves as an n type donor. The impurity oxide remains at the interface where it can build up a blocking layer. The interface resistance starts out at tens of ohms and increases over the life of the tube to as much as several thousand ohms.
Bell Laboratories really pinned this one down in their development program for ultra reliable tubes for their undersea cable networks. These special repeater tubes have lain undersea for 35 years now, or 300,000 hours, and they still work. Bell knows tubes.

sources, from which I have borrowed liberally:

Danforth, W. E., "Elements of Thermionics," Proc. IRE, vol. 39
pp. 485-499, May 1951.

Frost, H. B., "High-Purity Nickel Cathodes: Performance Studies,"
Bell Laboratories Record, pp. 18-22, January 1961.

Gewartowski, James W., and Watson, Hugh A., "Principles of Electron
Tubes," van Nostrand, 1965
Chapter 2, "Electron Emission," pp. 30-73.

Hampel, Clifford A., editor, "The Encylopedia of Electrochemistry,"
Rheinhold Publishing, 1964, "Work Function," pp. 1174-1176.

Hensley, E. B., "Thermionic Emission Constants and Their
Interpretation," Journal of Applied Physics, vol. 32,
pp. 301-308, 1961.

Holdaway, V. L., van Haste, W., and Walsh, E. J., "Electron Tubes
for the SD Submarine Cable System," Bell System Technical Journal,
vol. 43, pp. 1311-1338, July 1964.

Kern, H. E., "Research on Oxide-Coated Cathodes," Bell Laboratories
Record, pp. 451-456, December 1960.

Kohl, Walter H., "Handbook of Materials and Techniques for Vacuum
Devices," American Institute of Physics Press reprint 1995,
Isbn 1-56396-387-6 Chapter 16, "Cathodes and Heaters", pp 475-528

Papers cited here go into the rat tube paper collection. Copies of Kern and Frost will be sent to Ned for web publishing (at his discretion). Danforth and Hensley are much too technical. Holdaway is excellent and fascinating but too long to post on the web. Dang, now I have used up a couple of tube quiz questions.


steve bench:

Most interesting article Gary. Forced me to drag out a few of my physics texts to go over some of this. 

The partial quotes I included here relate to a couple of questions..... The work function of Ni (5.03eV) vs that of oxide coated Ni (1-2 V) implies a 3-4 volt potential difference within the cathode structure. Is this affecting the reactions to either enhance the replenishment of barium at the cathode surface (or deplete it) via electrochemical reaction instead of contamination? If this is true, it suggests that there may indeed be a valid method for regeneration of "old" tubes, assuming a reasonable maintenance of vacuum.

There seems to be a difference in tube curve characteristics attributed to thoriated tungsten -vs- oxide coated filamentary devices (the sonic preference for DHT?) Any comment about the physics of the difference between W / WTh / BaO-Ni emitters? It seems there is quite a bit of difference in the electrochemistry. The only reported effect I can readily recall is the difference in the saturation characteristics of the emission (temperature limited as well as voltage limited). I assume there is no field emission effect issues here, as they require much higher (10 exp 6 v/cm) fields than we're talking here.

Thoughts appreciated. Again, thanks for the nice article.
Best Regards,
gary longrie:


Back to me hey? So you want the answer to life, the universe and everything? Tricky. I will have to give it some deep thought.
Any answer I give now would just be off the cuff. I am not comfortable with the notion that contact potential volts are the same as power supply volts or battery volts. It's as though the voltage is there, as long as you do zero work with it. I need to explore more. We have a couple of graybeards that were here back during the heyday. And the place is swarming with hotshot semiconductor physicists too. They certainly know work functions and contact potentials.
As for the DHT question, I have lots of material but haven't actually done any digging. To a zeroeth approximation, the oxide cathode has a lot more series resistance than the thoriated tungsten cathode. Hmmm, thoriated tungsten. Might be another misnomer as tube tungsten parts are also carburized. Might not apply to filaments though. I'll find out. In the meantime, I recommend to you a copy of Kohl as part of your essential tube library. In addition to being comprehensive about this stuff, Kohl lists extensive references out in the hardcore physics and engineering journals. The cathode chapter has something like 215 citations. He doesn't just lump them at the end. Each time he introduces a new idea, paragraph by paragraph, he immediately references the appropriate citations. That way you know exactly where to go for more depth.

There will be time this weekend as I am staying off the highways.

regards and a safe holiday

Hi Gary,
I'll look that reference up. I found a suprising amount of material in an old college text "Electricity and Magnetism" by Arthur Kip, 1962, McGraw- Hill. Strange, I didn't have tooooo much use for it at the time, (bought it as backup for one of the physics courses), but I still find occasion to refer to it. Interestingly enough, it mentions ALL FOUR of your "kinds of emission" quiz question. Also covers the same relation between work function and contact potential as you mentioned in your post.

Interesting concept, that of the cathode resistance. That should show up as increased feedback over the lifetime of the tube. Producing, of course, an increase in plate resistance. Lo and behold, that happens.

I've seen texts listing the thoriated tungsten as being carburized and some stating the active region is a thin layer of thorium on the surface of the tungsten. Not sure which if either is really correct.

Thanks again, and have a nice holiday.

Best Regards,