• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Why Gold Grids

Status
Not open for further replies.
If we look at the ratio of emission current for two different materials at a constant temperature, we get exp(-q[deltaW]/kT). Plugging in a deltaW of 0.8eV (using Kohn's numbers) and T of 1000K, the ratio for gold to molybdenum (the usual grid wire) is about 0.0001.

If we want to quibble about the temperature and plug in 800K, the ratio is even smaller.

My measurements are not the great thing, but made me suspect about the Richardson's constant, it is material dependent, here most repeated values

Au

e φ = 5.1 eV,..... Ao = 14.7 A/(cm² ºK²)

Mo

e φ = 4.15 eV,..... Ao = 55 A/(cm² ºK²)

Then, at 1000 ºK

Jo(Au) ≈ 2.9 x 10⁻¹⁹ A/cm²

Jo(Mo) ≈ 6.7 x 10⁻¹⁴ A/cm²

Also

Jo(Au) / Jo(Mo) ≈ 4.3 x 10⁻⁶

So far these results favor Keit's argument, even with one more order of magnitude due to Schottky effect.
 
Last edited:
Work function is the key to electron emission and it quite well defined. It does vary a lot with alloy.........
Perhaps the work function of any Au/Ba or Au/BaO alloy is more the issue, since that might likely form the surface ?

One further thought, might Au et al act catalytically to breakdown of either BaO vapour or minority contaminants which would otherwise form an emissive layer ?
 
Perhaps the work function of any Au/Ba or Au/BaO alloy is more the issue, since that might likely form the surface ?

One further thought, might Au et al act catalytically to breakdown of either BaO vapour or minority contaminants which would otherwise form an emissive layer ?

In general, what gets vaporised from the cathode is the same stuff that provides the emission - free Ba atoms and some free Sr atoms.

If there is free Ba and/or free Sr, there must be free O as well. The O becomes part of the tube gas until it reaches the getter flash, whereby it gets captured as MgO (assuming magnesium gettering).

It isn't actually BaO and SrO that does the emitting.

In any BaO-SrO cathode, there's a tiny amount of atomic Ba and atomic Sr present throughout the matrix. Heat, and the process of electron emission, causes the Ba and Sr to evaporate (more correctly, "sublime") from the surface and become part of the tubes' gas until they land on something cooler than the cathode - the more cooler the better. Because it is one of the the coolest parts of an operating tube, most Ba & Sr free atoms end up in the getter flash, where they do no harm at all, or because of its enclosiveness, the anode, where no harm is done (especially if anodes are blackened by carbonising - carbon has similar emission suppressing effect to gold).

But, over time, a small fraction of vaporised Ba & Sr end up on the grid, which spells grid current and trouble.

Heat also allows the diffusion of more free Ba and Sr atoms from deep within the oxide layer, replacing what has been evaporated. Diffusion of Ni from the underlying cathode metal tube assists this.

Where a tube cathode suffers from ion bombardment (as when made to supply an anode current beyond the cathode's emission capability), entire BaO and SrO complexes or one or more molecules may be knocked out of the cathode, joining the tube's gas and looking for something relatively cool to settle on.
 
Last edited:
Ill effect to the cathode

If we look at the ratio of emission current for two different materials at a constant temperature, we get exp(-q[deltaW]/kT). Plugging in a deltaW of 0.8eV (using Kohn's numbers) and T of 1000K, the ratio for gold to molybdenum (the usual grid wire) is about 0.0001.

If we want to quibble about the temperature and plug in 800K, the ratio is even smaller.

Sy,

Do we know the temperature that the gold coated cathode is operating? I recall reading lately that the upper limit of the range is near 700 degrees to limit/prevent vaporization of the gold with ill effect to the cathode.

DT
 
Last edited:
Sy,

Do we know the temperature that the gold coated cathode is operating? I recall reading lately that the upper limit of the range is near 700 degrees to limit/prevent vaporization of the gold with ill effect to the cathode.

DT

Perhaps you mean grid, not cathode. Gold is useless as a cathode. It melts before you can get it hot enough to provide less than 1 nanoamp.

You can also forget about this nonsense about work function of gold being important. Unfortunately it appears this thread has been blighted by a certain person who hates been wrong - and as a moderator he should know better. The fact is, at around 1050 K nothing does any thermionic emission greater than nanoamps apart from oxide coatings - a fact you can check in any textbook on thermionic emission. At lower temperatures, you are even more comparing an extraodinarily tiny nothing with another extraodinarily tiny nothing.

The hottest part of the tube that a grid can "see" is the cathode. Oxide coated cathodes run at 1050K, so you can infer an upper limit of about 1000 K. The collest thing a grid can "see" is the anode. In a power tube anodes run at up to 750 K. Due to conduction out through the grid connecting lead, the grid temperature can be less than 750 K.

In Heat Transfer in Receiving Tubes, Schade O H, RCA 1961, he gives over many pages an explanation of how to calculate grid temperature. Read it, and I bet you decide the same as I did - there are more important things to do than calculate the temperature of grids! Such as doing the washing, painting the house.....

However, he does give the RCA 6L6G (which does not emply gold plated grids) as an example. Here's the data:-
Cathode................. 1050 K
Anode...................... 688 K
Control grid siderods....565 K
Screen grid siderods....788 K

The 6L6G data was obtained from a specially made custom 6L6G fitted internally with thermocouples. The tube was run continuously with anode 18 W, screen 1.25 W.

Note that, apart from the cathode, the hottest part of a 6L6G is the screen. You can see why larger power tubes have cooling wings welded to the screen grid siderods (and not to the control grid siderods).
 
Last edited:
If yiou plug the relevant constants into the Richardson-Dushman equation, including grid temperature in an oxide cathode tune (1000 K), the work function of various metals, Boltzmann's constant, A= 120 A/cm^2, you find that:-
a) the emision for various metals that could be used for grids varies over about an order of magnitude;
-but-
b) the emission is not even nanoamps.

Thus, unless oxide material migrates from the cathode to the grid (which it does by vaporisation) there cannot be a problem. Measures to reduce grid emision are thus directed to nuetralising the migrated oxide, or preventing it from landing on the grid.
The outcome would seem to depend critically on the work function of whatever grid surface metal/alloy might result after any migration of cathode material to the grid. The aim is to conserve very low grid emission after this process, not to fix any intrinsic emission from the raw grid material per se.

If the work function of the grid surface after any migration were low enough, then I calc it seems plausible that grid emission could be sufficient to be significant. Apparently, this does not happen in the presence of a thin gold coating for the grid. Depends on surface area of grid wire too, uniformity of any migration, and critically on both grid temperature and uniformity of temperature across the grid I suppose.
 
Perhaps the work function of any Au/Ba or Au/BaO alloy is more the issue, since that might likely form the surface ?

Entirely possible. That's why I want to do a reasonable calculation of the emission from metal electrodes to see if it falls anywhere near experimental numbers from new tubes. If not, then the issue is the work function of metal oxides and how they deposit and absorb. I don't think gold will act catalytically, given its inertness, but it can potentially absorb adsorbed species into its bulk.

Now, metal oxide work functions ARE hard to measure, not because of the instrumentation (which has no problem resolving 0.01eV differences), but because of the huge variability sample to sample and from spot to spot on a sample. Nonetheless, it's a much lower number than pure metals, which is why they are used as thermionic emitters.
 
I don't think gold will act catalytically, given its inertness, but it can potentially absorb adsorbed species into its bulk.
Certainly the gold is not acting catalytically.

While I was entirely correct in my earlier posts that the work function of gold has absolutely nothing to do with why it is chosen, my assertion that it is about keeping the grid clean of contamination needs some adjustment:

Rod Coleman cited Beck, who said "..the barium evaporated from the cathode diffuses into the gold instead of remaining on the surface."

In A Manual of Materials for Microwave Tubes, Thornburg D L, Thrall E S, & Brous J, RCA 1961, in says on page 63: "Gold internal surfcaes are preferred for tubes with oxide or matrix cathodes; evaporated barium will diffuse away from the surface into the metal".

Thus the virtue of gold is that barium atoms readily diffuse into it, where they cannot emit electrons, leaving the surface clean.

Now, metal oxide work functions ARE hard to measure, not because of the instrumentation (which has no problem resolving 0.01eV differences), but because of the huge variability sample to sample and from spot to spot on a sample. Nonetheless, it's a much lower number than pure metals, which is why they are used as thermionic emitters.

Sy seems to have some misconceptions here. It is not the oxide that emits electrons.

The manufacturing processes of vacuum tubes leave an amount of free (that is not chemically combined) Ba and Sr atoms diffused throughout the BaO-SrO grain matrix. It is the free Ba and Sr atoms that have access to the greater tube vacuum that do the emitting of electrons.

Heat enables electron emission; it also promotes the evaporation of surface free Sr and Ba (especially Ba due to higher vapour pressure of Ba). Cathode heat also replenishes atomic Sr & Ba by diffusion of same from deep within the oxide matrix.

If tubes are overloaded, the loss by evaporation can be greater than the diffusion replenishment - emission then falls. If a tube that has lost emisson in this way is run at a higher than normal heater voltage for a short time, emission is recovered.

Sy need not pester his students for a work function measurement. It is not the oxide that does the electron emitting. He can get the work function of Sr and Ba from standard tables. The values are:-

Ba........2.52 eV
Sr........2.59 eV

We might thus expect, in a cathode mix containg about equal proportions of BaO and SrO, that the cathode will have a work function of 2.55.

Real cathodes do somewhat better, appearing to have a work function lower than 2.55. Reasons include the following, in descending order of importance:-
1) Oxide cathodes have, on a microscopic scale, a very rough surface. The emission of electrons is similar in physics to the emission of atoms (evaporation). In both cases surface roughness improves emission.
2) Tube manufactuers discovered various proprietry ways of doctoring the carbonate mix to make it work better; they discovered that some metals when used for the cathode sleeve improved things (eg nickel);
3) Schottky effect slightly improves emission - because the voltage gradient within the cathode matrix accelerates electrons toward the surface, it reduces the amount of energy that has to be sourced from heat to ovecome the work function barrier. Popilin was right in principle, he was just wrong on the magnitude of the effect, which is small (few percent)


Actually, I think Sy has had a lend of us the whole time.

Now, metal oxide work functions ARE hard to measure, not because of the instrumentation but because of the huge variability sample to sample and from spot to spot on a sample.

Nah, mate, its hard to measure the work function of metal oxides like BaO and SrO because they are hopelessly unstable when not in a hard vacuum. To prepare a sample, you essentailly have to start with the carbonates and make a vaccum tube with all the care and process complexity, including phased bakeout while pumping, that tube makers employed. And then get rid of the free Ba and Sr atoms. Evaporation of Ba and/or Sr in a chamber containing only oxygen doesn't work - you get a complex mixture of compounds. Perhaps students measured TiO2, which is very stable and has a work function of 5.4 eV.
 
Last edited:
Docs from Western Electric for some of their small glass tubes say it reduces secondary emission from the grid. Fwiw. Can't cite, because I'm not sure where I read it, but of this I am pretty certain, would testify if asked? 😀

Also, I can tell you from my personal experience with a certain small glass WE tube used in audio circuits, the ones made earlier sound better (yes they do) and also the examination of broken ones (old vs. newer) shows that the gold plating got noticeably thinner as the production became more recent. Coincidence?

_-_-
 
Nah, mate, its hard to measure the work function of metal oxides like BaO and SrO because they are hopelessly unstable when not in a hard vacuum.

You do understand that PES measurements are done in hard vacuum? I am under somewhat of a disadvantage here, having actually done PES (edit: and KPM) in a top surface science lab.

Putting aside the irony of the reporting of work functions of metal oxides to three significant figures after denying that can be done, I would still like to work through the exercise of doing the basic maths before devolving into handwaving and pulling data out of context. Morgan Jones has kindly offered to do the physical grid measurements of some ECC83, since no-one seems to have volunteered hard data on that. From there, we can actually run through some more rigorous calculations to get a better idea of the mechanism of grid emission and how this might be affected by gold.
 
SY, I don't have any "modded" 12AX7 around unfortunately to measure grids. Only got 2 of them from an old Heathkit equipment.

----------------------------------------

Tubes would work so much better if the room temperature emitters had been solved back then:

"Physics World, October 2000, page 25 to 26 or Physical Review Letters 2000 Vol. 85 page 864.. Two scientists, Vu Thien Binh and Christophe Adessi, have solved the problem of a room temperature thermionic emitter 50 years too late. They use a 50 nm film of Titanium oxide on a nickel plate as a cathode. It has a mere 0.1 electron volt thermal barrier which is satisfied easily at near room temperature. The technique is robustly stable in the normal vacuum tube environment against adsorbed surface contaminants as well as ion bombardment. Previous attempts to achieve this using micron scale pointed field emitters were not robust or stable. Two construction techniques for making these new cathodes are suitable. One is vacuum sputtering, the other is sol-gel plating. The latter is similar to electro-plating thru a gel solution."

That was the 1st breakthrough back in year 2000. Now there are multi layer thin film cathodes that are even better. I guess the audio tube makers don't read the journals, or they are waiting for the patent to expire.
 
Last edited:
Putting aside the irony of the reporting of work functions of metal oxides to three significant figures after denying that can be done
I can't quite work out what that is about, posibly you have misread me.


I would still like to work through the exercise of doing the basic maths before devolving into handwaving and pulling data out of context. Morgan Jones has kindly offered to do the physical grid measurements of some ECC83, since no-one seems to have volunteered hard data on that. From there, we can actually run through some more rigorous calculations to get a better idea of the mechanism of grid emission and how this might be affected by gold.

I don't know the dimensional data for a 12AX7 or ECC83. Here is data for the framegrid (therefore a planar tube) WE 417A VHF triode as follows:-
Grid wire diameter............0.0074 mm
Grid pitch.......................0.065 mm
The ratio of grid surface area to cathode area is thus about 1 : 2.80. This is probably about the smallest ratio you'll find as this is a high gm frame grid tube.

I don't see why you are so keen to do an exact calculation though. There are so many orders of magnitude diffrence between pure metal emission and oxide cathode emission at typical tube temperatures, it simply doesn't matter what the grid area is any more than the work function of whatever metal reasonable to find employment as a grid you pick. You'll get maybe an order of magnitude variation out of each, but the big picture of oxide cathodes vs pure metal at 1000 K is many orders of magnitude.

Heck, you don't need to do any calcs. Just look up a graph of emission current density vs temperature for various metals and oxide cathodes in any good textbook on thermionic emission.
 
Last edited:
It's easy if you're a materials specialist. Schlenk techniques are something any chemist is familiar with, especially those of us who have done significant amounts of catalyst work. Standard stuff in any lab.

Blah blah blah blah.........

For those who might be curious: Schlenk techniques are air-free techniques (not hard vacuum cf vacuum tubes, but can have very good vacuum), typically employing a Schlenk flask.
 
Last edited:
I have experimental grid leakage data for 12AX7, so that's what I want to use for calculational comparison.

Question that I truly don't know: I've only been in one tube manufacturing facility, and it sure didn't look like they were handling the oxide-coated cathodes in a vacuum. Ladies on the assembly line had a pot of some oxide paint and applied it to the cathodes right at their workstations. Was this atypical?

not high vacuum

Under 1 micron, so yes, high vacuum. It helps to have actually done this for a living rather than using wikipedia. edit: some people are running under 10exp-10, but I didn't need to work that low.
 
Status
Not open for further replies.