Low cost hermetic seals

[TimFox]

I haven't dealt with Ceramaseal since graduate school (during the Nixon administration), but they were reasonable to deal with back then.  The indivicual feedthrus require welding.
see
https://www.ceramtec.com/ceramaseal/feedthroughs/

[SteveP]

Hi,

Regarding the statement that "epoxy is permeable to oxygen", people may not want to discard it as a solution quite so quickly. Based on some research I did a while back, the issue isn't *whether* epoxy is permeable, it is *how* permeable it is. For instance, if only one molecule of oxygen gets through per decade of use, then it would probably be deemed adequate as a barrier.

So just how permeable is epoxy to oxygen? I don't have the reference handy at the moment but the best answer I found was in a report issued by some researchers investigating the permeability of a large number of materials to various substances from the point of view of radioactive containment. The number they gave was 0.128 barrers. OK, what the heck is a "barrer"? And how do we use that information?

It is named after Professor R.M. Barrer who was, apparently, an early pioneer in the field of permeation and the barrer is the unit applied to the coefficient of permeability of a substance. This coefficient is given by the formula P = q*t/A*deltap where:

P is the permeability (in barrers),
q is the mass flux of the permeant (10E-10 cm**3/sec),
t is the thickness of the membrane (cm),
A is the area of the membrane (cm**2), and
deltap is the difference in pressure across the membrane (cmHg).

In other words, if our material has a permeability coefficient of 1 barrer we will get
10E-10 cm**3 of gas across the membrane, per second, if our membrane has an area of 1cm**2, is 1 cm thick, and we have a pressure difference of 1 cmHg (about 0.193 psi = 0.013 atm = 0.013 bar).

Since we want the contents of our container to last much longer than a second, let's do the calculation for a somewhat longer time period, say 100 years. That turns out to be about 3E9 seconds which, if we take the permeability of epoxy to oxygen as 0.128 barrers, gets us 0.128 * 10E-10 * 3E9 = .384 cm**3 of oxygen. At STP, the density of gaseous oxygen is about 0.00143g/cm**3, so we would get 0.00055 grams of oxygen in 100 years!

Since our membrane area (diameter of the hole in the container) isn't likely to be 1 cm**2, the thickness of the membrane (the thickness of the wall of the container) isn't likely to be 1 cm, and our deltap isn't likely to be 1 cmHg, let's do the calculation using more realistic numbers.

Take deltap: here's where things get a bit tricky. If we fill our chamber with dry nitrogen, it will presumably be at atmospheric pressure, as will the atmosphere on the outside of the container. So the pressure difference is zero, and there won't be any movement of oxygen across the membrane, right? Nope. Here, we need the concept of "partial pressure" which is simply the pressure not of the atmosphere as a whole, but of the *oxygen* in the atmosphere compared to the pressure of the oxygen on the other side of the membrane. Well, the pressure of the oxygen on the inside of the container is zero, since there is zero oxygen present (because we filled the container with nitrogen). On the outside of the container, the pressure of the oxygen is simply the percentage of oxygen present in the atmosphere (21%) times  the pressure of the atmosphere which is 1 atm, giving us an "oxygen pressure" of 0.21 atm and therefore a pressure difference (deltap) of 0.21 atm which is about 16 times the 0.013 cmHg of the barrer definition.

As for the membrane, let's make it easy on ourselves and consider a hole 1 mm in diameter through which our wire will pass and neglect the diameter of the wire itself. This will give us an "A" (area) of 0.785 mm**2 or 0.00785 cm**2. Let's further assume that our container's wall is 1 mm thick, giving us t = 1mm = 0.1 cm.

So now we have a deltap that is about 16 times greater than 1 cmHg, and an A/t of about 0.0785 resulting in a factor of about 16 * 0.0785 = 1.256; this means that for a hole of that size we would get about 0.00055 * 1.256 or about 0.0007g oxygen per 100 years.

You can take the math from here: ten such holes would result in 10x that amount of oxygen, if you only cared about a ten-year period it would be 1/10 of that amount, if you doubled the thickness of wall material it would be 1/2 that amount, and so on.

As it happens, manufacturers of LEDs care about sealing out oxygen and I ran across a paper that reported the results of sealing LEDs with epoxy. They used tritium as a tracer and found that the oxygen permeation was about 3x the above calculation for the type of epoxy they used. Epoxies differ, of course, so the exact number for something like JB Weld  would probably be slightly different, but I think that the calculation gets us in the right ballpark.

Given that some slight amount of oxygen will cross the membrane, we might simply "scavenge" it with an oxygen absorber. These are often used in packaged food products (and we've all seen moisture absorbers in electronics packages)--they're simply small sachets of a material designed to react with oxygen (or moisture) thereby removing it from the air. For oxygen, iron oxide is one such material (but it requires moisture to do its job, and we don't want moisture in our container any more than we want oxygen, so that's out) and ascorbic acid (vitamin C).

Based on what I've read, it takes one unit (mass) of ascorbic acid to absorb 2 units of oxygen, so even a single gram of ascorbic acid would be far more than we'd need. What I haven't been able to ascertain is whether any byproducts in the reaction (which is fairly complex) would have any effect on electrical components (solder, wire, etc.). I also don't quite understand whether moisture is required for the absorbing reaction to take place, so those two things would have to be investigated if one were to decide that 0.0007g of oxygen per century was intolerable. As noted above, in 1 cc of atmosphere (at STP) there is about .00143g of oxygen, so over the course of a century, with no oxygen absorber, the oxygen content inside the container would rise to about half the level of what it is in the normal atmosphere.

I'd be interested in the results if someone cared to take the idea of an oxygen absorber further and discover what, if anything, could be safely and effectively used around electronics.

Oh, before someone asks, the permeability of epoxy to moisture was listed as 0.013 barrers or about 1/10 that of oxygen.

Hope all this helps,
--Steve

[LaserSteve]

In light of the above post, I give you Torr Seal:

http://www.agilent.com/cs/library/datasheets/public/Data%20sheet%20TorrSeal3ok.pdf

and my beloved 1C Epoxy...

https://www.duniway.com/part/ep-8034

Both are "low outgassing" materials as well.


Steve

[richard.cs]

Given that some slight amount of oxygen will cross the membrane, we might simply "scavenge" it with an oxygen absorber. These are often used in packaged food products (and we've all seen moisture absorbers in electronics packages)--they're simply small sachets of a material designed to react with oxygen (or moisture) thereby removing it from the air. For oxygen, iron oxide is one such material (but it requires moisture to do its job, and we don't want moisture in our container any more than we want oxygen, so that's out) and ascorbic acid (vitamin C).

Based on what I've read, it takes one unit (mass) of ascorbic acid to absorb 2 units of oxygen, so even a single gram of ascorbic acid would be far more than we'd need. What I haven't been able to ascertain is whether any byproducts in the reaction (which is fairly complex) would have any effect on electrical components (solder, wire, etc.). I also don't quite understand whether moisture is required for the absorbing reaction to take place, so those two things would have to be investigated if one were to decide that 0.0007g of oxygen per century was intolerable. As noted above, in 1 cc of atmosphere (at STP) there is about .00143g of oxygen, so over the course of a century, with no oxygen absorber, the oxygen content inside the container would rise to about half the level of what it is in the normal atmosphere.

Thoughts on using a reactive metal as the absorber? I'm thinking magnesium ribbon or similar which will react with oxygen without water present and produce a solid oxide which should be reasonably easy to contain.

[LaserSteve]

OK, there are US Military Specifications for solderable feeds.  See this manufacturer's site:

http://www.shp-seals.com/aboutfeed.html

Steve

[SteveP]

Re the use of magnesium (or aluminum) as an oxygen absorber, the idea doesn't work because the oxidation creates an impermeable layer on the outside of the metal which halts further oxidation after the layer is only about 10 nanometers thick--far too soon to use up the quantity of oxygen we're talking about. There are, of course, more energetic metals one might consider--sodium and potassium, for instance--but they have the unfortunate side effect of bursting into flame upon exposure to the atmosphere.

--Steve

[Ian.M]

The addition of a small quantity of Mercury allows Aluminium to be rapidly oxidised by air at room temperature.  The Mercury dissolves some of the the Aluminium which then preferentially oxidises at the surface of the resulting amalgam liberating the liquid Mercury to dissolve more Aluminium.  As the Aluminium oxide formation causes the surface under it to form a liquid Mercury film, the normal stable, firmly attached Aluminium oxide film cant form so the reaction continues.   The Aluminium surface *MUST* be initially free of oxide as the Mercury cant break through the oxide film, but even a small fresh scratch is enough to start it off.

I would expect a similar reaction with Magnesium. 

However the use of Mercury is problematic, and even apart from the toxicity and environmental issues, the presence of Mercury vapour in the enclosure is likely to seriously affect long-term reliability as any condensed drops will alloy themselves with most metals if a metal surface is exposed.

[CatalinaWOW]

Now that you know how to figure out how much oxygen and water get across a given barrier you need to figure out how much will cause you a problem.  Even harder question.  Depends on material and construction of your resistors.

As an aside, why not eliminate the thermal EMFs by the classic technique of paired junctions?  In your application it should be relatively easy to keep the junction pairs at the same temperature.  Then you will open up your material selection problem.

[richard.cs]

I've certainly had bits of magnesium ribbon (pure, not an alloy with aluminium) turn completely into white dust after ~10 years in a corked test tube. Clearly there are some circstances in which it doesn't pasivate but it's possible that I had moisture ingress as well as oxygen.

Another option might be a finely powdered reactive metal, freshly milled alumiumium dust that's been kept away from air. Normally you'd introduce oxygen slowly after milling it to oxidise the surface without it catching fire but if you don't then there's alot of surface area there to absorb oxygen on before it passivates. Might be more work and danger than sodium, which is normally fairly safe and can be handled in air, small pieces usually don't catch fire in water either.

I just prefer a solid oxide to an unknown mix of organics which may break down into water later.

[SteveP]

The OP's requirements aren't super-clear to me even at this point, but apparently two of the objectives are (1) for a small number of wires to pass through to his resistors and (2) he wants to do this for less than $5 per wire. He stated that he wanted to guard against moisture and oxygen, but did not explain why.

In my own case, I was investigating putting an entire (small) board in an oven and didn't want the bulk and cost of a bunch of pass-thruough devices and simply wondered if wires/epoxy would do the trick. I had further noted that while some pass-through capacitors are made from glass/kovar, many are actually made with epoxy! (Remember, a lot of these are primarily aimed at keeping out RF rather than moisture.)

Regarding *oxygen*, as far as resistors go, modern precision resistors are, as I understand it, made from Evanohm which, being a nickel-based alloy, is basically impervious to oxygen. Manganin, on the other hand, is susceptable to surface oxidation, so the Thomas-type resistors made with it during the 1930s by the NBS (and others) were sealed (with solder) except for the opening through which the wires passed--that opening being sealed with silk impregnated with varnish. The air inside the Thomas resistors was not treated specially--it was not evacuated or replaced with dry nitrogen; they just made sure there was as little of it as possible and used large-diameter wire to reduce the surface area/volume ratio so that surface effects would be held to a minimum. Five of these resistors are still (or were as of a report issued in 2003) being used as working standards for the Ohm (the QHE device being used as the primary standard).

So I don't know why the OP was concerned with oxygen unless he knows his resistors are made with Manganin. If they are, then I suspect, based on the experience of the NBS, that replacing the air with dry nitrogen and simply sealing around small diameter wires with epoxy would be more than adequate. If they're made with Evanohm, then the concern with oxygen might be misplaced.

Dealing with *moisture* is far easier--simply replace the air with dry nitrogen, include a packet of silica gel which is commonly used with electronics today (and as far as anyone knows, does not turn to mush over time), use epoxy to seal the openings around the wires, and call it a day. For that matter, if the volume is small enough, you might even skip the dry nitrogen step.

Back on page 1, the OP expressed a concern that it wouldn't be possible to know whether the openings were actually sealed if epoxy were used. I can assure him that I have done this dozens of times for non-electronics related projects and the seals have proven gas-tight at pressures up to around 1,000 psi for tens of seconds. A thick epoxy like JB Weld (don't use laminating epoxy--it's not viscous enough) still flows well enough that if you coat a small section of the wire with an excess of epoxy and spin it slowly as you feed it through the opening, the epoxy will flow well enough to completely seal the opening. If the container has very thin walls, then simply surround the wire with a blob of epoxy on each side of the wall.

(I have mentioned JB Weld a couple of times; I do so merely because I've used it on many occasions over the years and find it to be reliable, relatively inexpensive, and widely available. Other brands may be better (or worse).)

--Steve
 

[Edwin G. Pettis]

Or you could do it the easiest way, just buy resistors that aren't sensitive to humidity, such as in the Ultra Precision Reference LTZ1000 thread, look for Ultrohm Plus resistors.  They are way cheaper than Vishay hermetics.

[splin]

The OP's requirements aren't super-clear to me even at this point, but apparently two of the objectives are (1) for a small number of wires to pass through to his resistors and (2) he wants to do this for less than $5 per wire. He stated that he wanted to guard against moisture and oxygen, but did not explain why.

Apologies - I should have made it clearer. I have a fair number of Vishay Beyschlag UXB 0207 precision thin film resistors that I acquired at very low cost. They don’t specify the materials in the datasheet but the stability spec is very good by thin film standards:
1000h     < .02%
8000h     < .04%
225000h < .12%

Ok, thats way worse than wire-wound resistors but mine are old stock so should be drifting much less by now. Much more significantly, I assume that the primary cause of long term drift is corrosion so I’d like to try putting them in a gas-tight hermetic enclosure to exclude oxygen and water vapour. They were also cheap enough to allow me to use multiple resistors in serial/parallel combinations to give some statistical improvement in the temperature and time drift characteristics.

I also have some Vishay foil S102K resistors which I also would like to try enclosing. They aren’t especially stable due to moisture absorption by the plastic encapsulant; I’d hope they behave more like the hermetically sealed versions which are very stable (spec’d to be <2ppm in 6 years), but very expensive.

This may well turn out to be a complete waste of time but it won’t have cost me much given that I have ordered 50 of the ex Russian military seals for < £6 including delivery.

So I don't know why the OP was concerned with oxygen unless he knows his resistors are made with Manganin. If they are, then I suspect, based on the experience of the NBS, that replacing the air with dry nitrogen and simply sealing around small diameter wires with epoxy would be more than adequate. If they're made with Evanohm, then the concern with oxygen might be misplaced.

Epoxy probably would be good enough but those Ebay glass seals are cheap enough that it’s not worth messing about with home made seals. Of course they might turn out to be faulty/rejects so I might try a simple pressure test. You’re right that if I had high quality wirewound resistors then I wouldn’t bother putting them in sealed environment given Edwin Pettis’s statements elsewhere that they are minimally affected by humidity or corrosion.

I have to say thank you to everybody who have contributed to this thread with way more high quality responses than I expected - especially LaserSteve and SteveP for his excellent analysis on calculating the permeability of epoxy – it’s one thing knowing the basic theory but getting real numbers and formulae for these relatively obscure topics is what makes these fora so wonderful.

Or you could do it the easiest way, just buy resistors that aren't sensitive to humidity, such as in the Ultra Precision Reference LTZ1000 thread, look for Ultrohm Plus resistors.  They are way cheaper than Vishay hermetics.

True, but the drift spec for the Vishay hermetic, < 2ppm/6 years, is quite a lot better than for yours at 2 (or3?) ppm/year. I’ve love to see some real-world measurements though as I wouldn’t be surprised if the stability of your resistors improve considerably over time (in a benign environment).

Actually I do intend to order some of your resistors but I’m having trouble deciding exactly how many and which values to order. 

[calexanian]

So this a year late, but......

I was told by an old timer that when they needed a quick hermetic seal for wide voltage and current ranges spark plugs were commonly used. They can be hard soldered or brazed. External connections can be made with non resistor plug wires, and you can grind off the ground tang and tack weld directly to the electrode. Best of all they are cheap as!

[ChristofferB]

Spark plugs are also better than pretty good in ultra high vacuum setups, according to "Building scientific apparatus" by Moore, Davis and Coplan. If you have some large stainless manifold with a nice thick blank to thread and mount them in, but they are pretty damn bulky for enclosure mounting!