cooling of objects in vacuum chambers

what

by

e to

rise

on

rer,

When

ink

we

nt

My

e in

to

from

"gut

lly

If my memory what I was told about the process of pumping down electron mic roscope chambers is anything to go by, getting the gas out happens fast and early, and the process that took the time was getting rid of the physicall y adsorbed water.

talks about the difference between multi-layer adsorbtion and the adsorbtio n of the final monolayer, where the bonding to the surface tends to be rath er stronger.

There's not enough energy involved to make much difference to the temperatu re of the adsorbing surface - the layer is only one molecule thick where it exists at all - but the equilibrium vapour pressure of water above that re sidual final monolayer is quite a lot lower than above a layer several wate r molecules deep.

We had hydrocarbon monolayers as well - where we hit them with an electron beam they'd turn into a thick (and eventually visible) insulating layer of polymerised hydrocarbon, unless the substrate was thin enough to be warmed by the electron beam, where you'd get a much less inconvenient layer of ele ctrically conductive gunge.

How are 'they' stored prior to vaccume?

Can you flood the chamber with nitrogen, then pump it down again with the same results?

Seems like typical out gassing of whatever, moisture etc...

Cheers

I don't think so! Not the air in most places anyway.

Just in room air.

Yes, we do this at experimental sites that have the nitrogen available. Definitely makes the pump-down to high vacuum go much faster. Never thought to see whether the detector cooling effect is different with a nitrogen fill. When we change targets or have to do some similar work in the chamber, we want to get it done and get back to beam on target and taking data as fast as we can, so we vent with nitrogen specifically to speed the following pump-down.

Jon

OK, I did the experiment, and there really is significant cooling. I propped a piece of thin aluminum sheet in the chamber, and taped an AD590 temperature sensor to the plate, with a little dab of thermal compound under it. I was easily able to show a temperature drop of 5 C or more while it was being pumped down. So, that maybe does support my theory. The detector in question was surrounded by a housing and a plate to protect it, so it was kind of enclosed in a Kelvin shield, that may have increased the effect. Here, I had a plate just propped in the chamber with no attempt at thermal isolation, and the sensor produces a tiny amount of heat, a couple mW. Well, coupled to that big plate, that heat ought to be negligible, but maybe once it gets down to real vacuum, it affects the measurement.

Jon

hat

der

Adsorbed water was the time determining factor is pumping down electron mic roscope scanning chambers. Pump-down time was pretty much determined by rel ative humidity, and it was worth filling the chamber with dry nitrogen when you wanted to open it to get quick pump-down after you'd closed it again.

For anything with any thermal mass, the latent heat of vapourisation of the adsorbed water isn't significant. Aluminium foil doesn't have much thermal mass.

Yes, we are well aware of this phenomenon, and use dry nitrogen to vent chambers when we are running an experiment at an accelerator lab. Even when we will pop the whole chamber open, it helps. When we only have to change a target through a small hole, it makes the pump-down several times faster.

Well, I used a piece of .031" aluminum, as it was a bit close to the 1mm thick silicon. But, I left the sensor powered up the whole time. I'm going to do it again but just turn the sensor on long enough to get a reading. I observed a temperature rise of at least 1 C per minute with the sensor on, and I'm guessing that was the self-heating of the sensor, at a couple mW. Insignificant in air, becomes more of a problem in vacuum.

Anyway, I think the adsorbed water effect is separate from the adiabatic cooling, and I think that is what I was seeing.

Jon

Great! (A little data can certainly help.) I was going to say a diode makes a nice sensor, ~10uA..-100uA.. run it hotter to mimic the load of the leakage in big detector, maybe.

(Aside: The whole subject of cooling by gas expansion is something that still confuses me (somewhat) in the details.)

George H.

I have used the AD590J sensor for years. it is an analog current source, with current proportional to temperature. 1 uA/C, referenced to absolute zero. So, 293 uA at room temperature.

Compress a gas, and it gets hot (feel an air compressor's tank, for instance.) Expand a gas, and it gets cold. Now, if we have gas in a chamber and suck some of the gas out, it gets cold. (See reports of sudden decompression in aircraft, they report the air turns white for a few seconds, it must get cold enough to go well below the dew point.) With no conduction to other objects, if you pump the air down to half an atmosphere pressure, I think the gas law says it will cool half the way down to absolute zero. Now, as the air density drops, its ability to absorb heat from objects in the chamber drops, also. Since the air in the chamber only weighs a gram or two, its ability to really cool objects is quite limited.

Anyway, I'm guessing if I leave the temp sensor off except when making a measurement, I think I will see about a 7-8 C drop with the setup I used Friday. I already showed a 5 C drop with the sensor left on.

Jon

Yeah, Well you've got adiabatic expansion, where the gas does work and is cooled.

And then you've got free expansion, (see article on Joule expansion for more details.)

But then as you know, if you take a tank of gas, crack the valve and let out the gas, the tank gets cool. (I've done a similar thing bleeding one (full) tank into and empty tank. The full tank gets cool and the (formerly) empty tank gets hot. I've never seen a detailed explanation of this process, (I haven't looked very hard either.) nor any numbers on how much colder/ hotter the tanks get.

George H.

(See reports of sudden

Air does not expand in a vacuum chamber; the volume is constant. Without expansion there is no adiabatic cooling. Have you actually observed the air temperature in the chamber cooling as the vacuum pump is applied?

Yes. Pumping down from ambient makes the sides cold to the touch. The effect reduces the lower the remaining air density, of course. But it will definitely have an effect on the detector.

Jon's original question is how to calculate the cooling effect on the device of residual air while it's still being pumped down. That will depend on whether its moving (circulating, etc) or stationary, but I'm not surprised at the effect he reports.

Clifford Heath.

'I do not know, how much this applies to silicon, but at least for porous materials, such as the PCB, this may have some effects.

In normal atmospheric pressure, there is a considerable

this external force is also removed. This may cause expansion in some materials.

The long (8 h) time constant sounds like some outgassing effect.

Couldn't it just be slow equalizing by black-body radiation?

One thing to try might be to put a temperature sensor in with a similar thin planar geometry to measure directly what is happening to the temperature rather than try to calculate it.

My instinct is that is that is a bit too slow for equalisation by black body radiation and that the large surface area is slowly losing adsorbed gasses to the vacuum. The chamber surface is essentially soon at ambient temperature and 4pi steradians so you can compute how much flux it is capable of supplying as a function of temperature difference to your device.

The curve of the stabilisation process may hold clues as to which process is actually the cause of the drift. My money would be on some form of outgassing from the large surface area but I could be wrong.

We always tried to provide a nice aluminium heatsink path to the stainless steel case but our problem was more thermal drift of the amplifiers that could be signal dependent if there was a large ion beam and insufficient cooling. We discovered some curious noise behaviour of certain opamps when actively cooled (some patents from the 1990's).

We were always suck with needing hard bakeable vacuum so the things you could use by way of materials were very limited.

The volume of the chamber is constant but by pulling the air out of it rapidly you are in effect making the chamber bigger as the pressure falls. Opposite of a bike pump making air inside hotter.

There is no doubt that the chamber will chill as you pump the air out.

The air DOES expand if you start out at atmospheric pressure and then pump down to vacuum.

I put a thin aluminum plate with a temperature sensor attached in the vacuum chamber and pumped it down from atmospheric to high vacuum. I think I described the ruslts of this test in a reply to my original message a couple years ago. I believe I recorded about a 5 C drop during the first part of the pumping, and then the temperature stayed low for a number of minutes.

This question was prompted by a pronounced rise in leakage current in large- area (100 mm x 100 mm square) silicon particle detectors after the chamber was evacuated. It ocurred to me that these could be cooled by adiabatic cooling, and then take a long time to warm the center of the detector via thermal conduction from the edges. So, I built a test setup and showed the effect was actually happening. The object needs to have a LOT of area relative to the mass to see any effect. The silicon detectors are REALOLY thin (100 um and less).

Jon