Vacuum-dehydrating hygroscopic components...

Stickman393

Stickman393

Well-known member
So.

I found a killer deal on some NPN germaniums. Got a good little batch, and wouldn't ya know it? They're all pretty well P-H-U-C-K-E-D. Leakage in the 2ma range. Meh.

But...hey. these are vintage. From about 1969 or so. Probably not perfectly stored over the years.

So I'm curious...I've read that moisture can cause these little fuckers to go leaky. I dunno. Let's experiment!

Well, if there's something that I'm well equipped to do: it's remove moisture from shit. HVAC mechanic, away!!!

So this. I set this up. We'll see.

First shot here is gonna be pretty "vibes" based. Most of what I've got in there is about 300uA of leakage. Imma let this thing go overnight and see what happens.

1000009473.jpg

1000009472.jpg
 
This seems like an interesting experiment, but unless you're going to find a way to "seal" the transistors after removing moisture, isn't it kind of pointless once they're installed in a circuit?
 
owlexifry said:
never considered that moisture could get inside these lil guys.

what about silica beads?
Click to expand...
Silica beads work too...but *much* more slowly.

Vacuum dehydration has a very specific advantage: boiling points are directly affected by pressure. Silica can only absorb water vapor that makes its way into the air at atmospheric pressure. I've probably got my chamber down around 1000 microns or so right now...though I haven't hooked up a vacuum gauge. That's on the job site right now.

But...1000 microns will get water's boiling point down to about 1⁰F.

Now...that's only half the battle. You still need to add the latent heat of vaporization in order to get that water to become a gas that can easily be pulled out of the vacuum chamber. Though, what that really means is that any water in the vessel will drop in temperature down to its boiling point until it absorbs enough heat from its surroundings to vaporize. Much like boiling a pot of water keeps the temp at 212⁰, but instead of adding heat on that case, we're letting the ambient heat do the work here.


Brett said:
This seems like an interesting experiment, but unless you're going to find a way to "seal" the transistors after removing moisture, isn't it kind of pointless once they're installed in a circuit?
Click to expand...

That all depends!

How much moisture is absorbed is a function of how much moisture is available to be absorbed.

A sealed guitar stomp box isn't particularly leaky. Its not watertight, but it's certainly sheltered enough to protect the components from the worst that the world has to offer. In truth, it would be fairly easy to keep the environment inside the pedal fairly dry simply by taping a desiccant packet inside the lid.

Time is an important factor as well: hygroscopic materials gradually absorb moisture from their surroundings, but how quickly they do so will vary on both the properties of the material as well as the hygrometric conditions in which it exists.

So far? This looks promising. Behold!:

1000009474.jpg

That little bit of cloudiness in the vacuum pump oil? That's moisture. I was running this thing for maybe 30 minutes before it turned that slightly milky color.

If this was a line set for a DX split system, I'd be calling my pipefitter right about now and asking why it looked like somebody peed in my refrigerant pipes.
 
Last edited:
Same^^^
Whiskers are also a thing. Internal crystal lattice type structures that grow to/from the casing and the P or N material. I forget. It perplexed transistor engineers for a bit.
Though I'm not sure if it increased leakage or just killed the functionality
 
iamjackslackof said:
This is an amazing experiment, and i can't wait to see how it turns out.

Could you coat the transistor bodies somehow to keep moisture out? Seems like even a thin coat of wax would help.
Click to expand...
Probably could, though then you've got the opposite problem should water make its way back inside. Whatever makes it harder for water to infiltrate will also make it more difficult to remove.

So: first shot? Maybe 4 hours on the pump, no real drop in leakage current. As far as I can tell. My vacuum pump oil was saturated with moisture again though when I just went out, so I turned it off and we'll see how it holds in the morning.

Got me looking into other remedies, like for tin whiskers. I'll see if I can find continuity between the case and leads tomorrow. Maybe a little current will knock em down if I find em.


iamjackslackof said:
Also I was convinced the post title was randomly generated and would totally be spam, but turns out it's totally accurate! :)
Click to expand...

I am a product. Of my environment. And also to buy. Buy me please. Send money.
 
I'm not saying that the experiment can't work, rather that the analysis needs some refinement. Let's assume that somehow the hermetic seals on the devices are somehow compromised, otherwise all is for naught. Then yes, germanium oxide is water soluble. However, for absorbed water to impact the device performance implies that the water has somehow modified the germanium below the surface of the transistor, implying chemical reaction, chemi-sorption, adsorption, etc (i.e., not just sitting on top of the surface waiting to evaporate). If water is chemi-sorbed or otherwise bonded to the germanium, then the pressure-dependent boiling point is not so relevant - the relevant factor then becomes the binding energy ΔE of the water molecule to the germanium, and it's release is now dependent exponentially on -ΔE/kT. If it has strongly reacted, any dependence pressure is likely to be weak.

Additionally, 100 microns is not particularly high vacuum for this purpose, as the mean free path for gas molecules is very roughly 1mm at that pressure (the average distance a gas molecule goes before scattering off another gas molecule, the actual # depends somewhat upon the specific molecule being considered). At 10^-6 torr pressure (~10^-3 micron) the surfaces under vacuum are bombarded by the equivalent of ~3 monolayers of gas every second (if all the gas atoms stuck to the surface). So at 100 microns, they are seeing the equivalent of 3x10^+5 monolayers of gas molecules every second, i.e., a lot. Together what this means is that as a water molecule tries to escape from the surface (i.e., "jump over" the binding energy barrier ΔE), it has a high probability of being knocked back onto the surface and being again chemi-sorbed.
 
Last edited:
I think the takeaway is that once a component gets p-h-u-c-k-e-d, it usually stays so.
 
God DAMMIT darwin. With your "theory" of evolution. Tell me: it you're so smart, how come it isn't a law?

Everybody knows that life is actually made of sticks. WHAT? PROVE ME WRONG!

*Ahem*. Uh...where was I?

Ah. Yeah.

So, I appreciate the input. I'm not fully studied on the actual mathematical side of things like gas density in varying levels of vacuum: my industry tends to consider a standing vacuum of 500 microns "good enough". We still have a large contingent of folks that can't even get that right.

Take this for what it is: a stickman in a stick world trying to make sense of things by banging rocks together.

So...yeah. I mean, I understand what you're saying regarding chemical reaction. The same thing happens with refrigerant oil when it's saturated with water: its damned difficult if not impossible to dehydrate at that point. We tend to have to flush out as much as we can from the system and replace it when that happens (like, when a water source heat exchanger develops a leak).

Your point regarding hermetic seals is well taken, though I can't help but wonder exactly how good those seals were in early transistor manufacturing.

My concern is with leakage, which (to my mind) could potentially be impacted by surface-layer water: it doesn't necessarily need to change the operational characteristics or the transistor: it only needs to provide an alternative path between the emitter and collector. I mean: tell me if I'm off base here (hehehehe, get it?).

I'm curious...you seem to have a firm handle on this sort of thing. At what point would a vacuum be deep enough to be useful in achieving dehydration at a chemical level? Is it possible?

I ask as 100 microns is by no means as deep as I can go. I originally quoted 1000 microns because this is a new setup that I haven't tested yet: thats just a guesstimate. I don't trust the gauge on the tank to make any pronouncements of deep vacuum since it's a full range device, and the hose is a crappy 1/4" charging hose.

I can hit 100 microns on this thing without breaking a sweat. That's babytown-frolics type shit. I'm confident I can take this guy down to 15-20 microns if I put in a little effort in improving the seals and use my *real* setup:

1000009476.jpg 1000009475.jpg

That guy's a 10 ton heat recovery VRF split system. Newly installed. I let it stand isolated from the pump for about two hours and it's only crept up 9 microns over that period. Not bad for HVAC.

That gauge is rated to +/-5% of reading at full scale and +/-5 microns. It should be fairly accurate.
 
You must log in or register to reply here.
Back
Top