cooling of objects in vacuum chambers

Sorry about the subject line, I was trying to be concise and cound't find better wording.

We have noticed a phenomenon when pumping down silicon particle detectors in vacuum chambers, and I had a possible explanation.

So, we put large-area silicon strip detectors in a vacuum chanber, and then notice the leakage current when reverse biased goes up for a CONSIDERABLE time, but eventually levels off. The last detector was the biggest we've ever used, about 200 mm diameter, and 1 mm thick.

My theory is that this thing has an insane amount of surface area and very little mass. Also, it has really poor thermal conductivity to the outside world. It is mounted on a piece of PC board material that is then mounted to an inner metal box in the vacuuum chamber. So, as the air is pumped out of the chamber, the air gets cold. Massive objects are not going to cool much, but the things with the greatest surface area/mass are going to cool the most. Then, the detector is going to warm up at a very slow rate. So slow, in fact, that the leakage only stabilizes after more than 8 hours.

Anybody have any experience with this? I know one could calculate the adiabatic cooling due to expansion of the air, but I have NO IDEA how one would calculate the cooling of objects in the chamber as the air is pumped out. It seems there would be some worst-case rate of pumping. Pump really fast and the air is all gone before much heat has left objects, pump REALLY slowly, and the air all maintains thermal equilibrium with the chamber walls. Somewhere in the middle of those extremes is where the most heat is drawn out of objects.

And, of course, the adiabatic cooling heads toward absolute zero as the pressure drops to hard vacuum levels, so that is not a useful point, either.

Thanks in advance for any comments, crazy or experienced!

(I did an experiment about 20 years ago with thermal conductivity in vacuum between hard surfaces without any conduction medium like thermal paste, and found that it is essentially zero! I had a piece of beryllium oxide heat conductor with a water loop thermal epoxied to it, in a vacuum chamber. I clamped an AD 590J temp sensor to it, and turned on the water cooling loop. The temp of the sensor changed a few degrees over tens of minutes. Then, I did it with one tiny drop of thermal grease, and the sensor reached equilibrium in less than 10 seconds. In air, with no grease, it took less than 30 seconds.)

Jon

Reply to
Jon Elson
Loading thread data ...

in

en

y
e
d

ut

l
o

d

ly

LY

is

er.

um

nd

I

p.

I
s

The thermal conductivity of gases is a bit weird. It's pressure independent until the men free path of the gas molecules becomes longer than the width of the gap over which you are measuring the thermal conductivity, then dro ps in proportion to pressure. It all makes perfect sense if you think about the heat being transferred by individual gas molecules, but it is counter- intuitive.

The electron beam testers I worked on a Cambridge Instruments used finely f ocused electron beam to scan the surface of working integrated circuit to m easure the voltage at the point scanned - by chopping the beam we made the images stroboscopic, so you could measure the voltage over a brief period - down to half a nanosecond - making the device the world's bulkiest and mos t expensive oscilloscope probe.

The integrated circuit had to be unpackaged, or at least in an unlidded pac kage, and it had to be running under vacuum for the electron beam to actual ly get to the surface.

When we looked at an ECL memory chip - which ran hot, even in air - we had to squeeze a length of copper braid under the package, and bolt the other e nd of the braid onto the metal wall of the vacuum chamber to keep the chip cool enough to work. A bit of vacuum grease at either end of the braid did help.

--
Bill Sloman, Sydney
Reply to
bill.sloman

On Tue, 10 May 2016 17:47:56 -0700 (PDT), snipped-for-privacy@ieee.org Gave us:

You never told us about your gayness before.

Reply to
DecadentLinuxUserNumeroUno

Unsurprisingly. That was a typo for "mean free path" as would have been obvious from the context to anybody who knew about ideal gas theory. Many don't.

--
Bill Sloman, Sydney
Reply to
bill.sloman

On Tue, 10 May 2016 18:28:14 -0700 (PDT), snipped-for-privacy@ieee.org Gave us:

I knew what it was, lattice boy. Don't need no primers or accusations of lack of knowledge from you.

Reply to
DecadentLinuxUserNumeroUno

Chronic troll a.k.a. "AlwaysWrong"...

--
DecadentLinuxUserNumeroUno  wrote in news:hn05jblbat0b314r631h8q1sk23v14hpl9 4ax.com: 

> Path: eternal-september.org!mx02.eternal-september.org!feeder.eternal-september.org!aioe.org!.POSTED!not-for-mail 
 Click to see the full signature
Reply to
John Doe

You've just demonstrated a lack of knowledge in your enthusiasm for getting into a lather about a typo. It's not knowledge you lack, so much as a sense of proportion.

Had you actually known you would have produced something more patronising, like "Surely you meant *mean* free path?".

Pity about your pretensions - you do make the shallowness of your claims to sophistication a bit too obvious.

--
Bill Sloman, Sydney
Reply to
bill.sloman

If you were more polite, people would not relish pointing out your typo's.

Dan

Reply to
dcaster

een obvious from the context to anybody who knew about ideal gas theory. Ma ny don't.

ons

tting into a lather about a typo. It's not knowledge you lack, so much as a sense of proportion.

ng, like "Surely you meant *mean* free path?".

s to sophistication a bit too obvious.

.

That's a hypothesis, but you probably need to take into account that there are lots of different people posting here, and some will point out every la st typo in the output of the most polite poster. It seems to have more to d o with how tight their knickers are at the time.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Hi Jon, interesting question. (too bad the flame wars seem to have found it already.)

I don't know silicon strip detectors. (first google hit.)

formatting link

So kinda a photodiode with lots of metal strips.

Does the leakage goes down if it is cooled? (that doesn't seem right to me.) Can you forward bias it and use the voltage drop (at known current) to measure the temperature of the device? (or stick a thermal couple on it?)

Hmm... does leakage return to "normal" levels after 8 hours?

Could it be picking up more ions in the gas, which now have a longer mean free path and can so have a better chance of hitting it. (If you can't tell i'm making things up.)

huh, OK I'm not sure how to calculate the cooling.... temperature of the gas I'm assuming this is only at lower pressure's once you've "roughed" out the chamber to ~100 mTorr or so? But I think I could take a stab at calculating the cooling. Given a gas at a different temperature than a small object. (find number of atoms/molecules hitting per second, assume atom arrives with average (gas) temperature and leaves with temperature of object... calculate energy loss per collision... at that would give a cooling rate.

It seems there would be some worst-case rate of pumping. Pump really

Before you go chasing the thermal idea, I'd like to be sure that it really is a temperature effect... perhaps you've already done that.

What's the magnitude of the leakage... maybe you are seeing more ions from cosmic rays?

George H.

Reply to
George Herold

Well, a GIANT, unsupported photodiode, yes.

Sure it does, just like any semi. Remember, these things have huge area, and we bias the whole detector, although it is broken up into thousands of pixels, they are all biased in parallel. The big round one we just set up was leaking 2 uA at 100 V reverse bias.

Hmmm, that's a very interesting idea. This is a $15,000 detector, so they may not want me to fool around with it too much.

We don't actually know what "normal" is supposed to be. What we did see was that leakage increased for a long time, then reached a plateau.

We get down to really serious vacuum in 20 minutes or so, better than 5 x

10^^-5 Torr. The Alpha particles have no problem flying a few inches at that pressure. Also, the current from the Alpha particles is too small to measure at the power supply. It resolves something like 10 nA in the least significat digit.

Oh, we are now sure it is indeed temperature-related. The 8-hour asymptote was kind of the giveaway.

Starts at 1.8 uA when we get good enough vacuum to turn on bias without risk of corona. Ends at 2.1 uA or so at thermal equilibrium. Cosmic ray leakage is absolutely TINY. If it was a serious problem, we'd never see the Alpha particles from our source. Even the Alpha particles cause a tiny current, must be a couple nA maximum with a strong source. If they run this detector without a mask plate in front of it, I might try to see if I can get a reading, but I doubt it. (The mask plate has small holes equal to about 20% of the area, to cut down the particle rate.) Generally, the only way to see the current change suddely is by letting light into the vacuum chamber.

Jon

Reply to
Jon Elson

Ughh. did I write that? ... I meant to write something else. Doesn't matter, I mis-understood your mechanism. Cooling and then a slow return to steady state. Hotter = more leakage. You've also got ~2 uA @100 V, 200 uW of power, could it be heating from that? (I can think of experiments, but a few hour time constant makes them all painful.)

Huh, so if I'm reading you correctly you are trying to see nA (or lower) of signal in the presence of 2uA of leakage.. and the change in leakage is a pain. Seems like it would pay to try and cool the detector. (If you could measure the Temp a control loop maybe?) A big hunk of cooled (LN2?) metal behind the detector and let radiation do it's thing? (radiation, as in radiative cooling.)

George H.

Reply to
George Herold

I really doubt 200 uW could heat a disc 8" in diameter by more than 1 C. That is a lot of surface area to radiate from, about 100 in^^2. 300 cm^^2, counting both sides.

Yes, with this long time constant, we SHOULD see a rise in current, if the bias was heating things. But, we DON'T see that, only after the pump-down. So, I think the 200 uW heating is NOT actually producing any measurable heating - which I would NOT expect.

Well, this thing has 128 conductive rings on one side, 128 pie wedges on the other. So, each of these rings or pies sees 1/128 of the total bias current. We are actually seeing signal currents in the fA range, I think. We get away with this as the ASIC that looks at these signals is tuned to the response of a particle going through the Silicon, and it tries to extract the signal and filter out the noise. These events last a couple hundred ns. We are on our 5th generation of the ASIC chips.

Jon

Reply to
Jon Elson

There might still be other possibilities; air has a few percent of Neon, which could diffuse into the silicon, and slowly diffuse out. It could be that you have an unintended dopant...

Reply to
whit3rd

If there is any moisture in there, when the pressure goes down, that moisture EVAPORATES, and that absorbs a lot of heat.

Mark

Reply to
makolber

A tiny amount of moisture might adhere to the surface, but it is glass passivated, so the moisture won't get into the Silicon. (Other stuff in the chamber could absorb more moisture, but it is not able to cool the Silicon.

Jon

Reply to
Jon Elson

Well, this detector is glass passivated, and 1 mm thick. Would Neon cause a reduction of reverse-biased leakage? I can imagine Helium diffusing in, but I have to doubt that Neon would be a major effect.

Jon

Reply to
Jon Elson

Hmm, (It didn't register how BIG it is.) (500W/ m^2 at 300K is ~15 W for 300 cm^2, 301 K would be ~15.2 W {15*(301/300)^4} Well unless you've got very very shinny walls. :^)

Oh so it's not really a problem, you are mostly trying to understand what causes the apparent cooling.

George H.

Reply to
George Herold

Well, **MY** theory is, it is cooling.

We have seen this phenomena a number of times before, and were puzzled by it. There is a well-known phenomena in radiation detectors where damage to the crystal lattice and implaning of ions causes the leakage current to rise until the detector is no longer usable. There is also another phenomenon where the leakage rises, and we send the detector back to the manufacturer, who removes the wire bonds, cleans the detector and then re-bonds it. When we get it back, the leakage current has gone down significantly. We think this is due to contamination of the surface with vacuum pump oils, but we are guessing.

In many cases, this rise in leakage current after pumping the chamber is muddled by hitting the detector with a lot of particles, so we can't distinguish radiation damage from other effects.

This time, we had a brand-new detector in a relatively benign environment for several days, so we could actually see this effect all by itself. My boss was quite concerned until we saw there was an asymptote to the rise in current. The huge size of this detector may have made this a more pronounced effect than we've seen before.

Yes, so now that we have observed this in relative isolation, we'd like to understand what went on. We do this stuff all the time, and if we could KNOW that every time we pump out the chamber, we'd see, for instance, a 20% decrease in leakage current that would then go back to baseline over the next 8 hours, that would be a real help in telling the normal behavior from the abnormal.

I've convinced myself that I understand the mechanism, but this is all "gut feel" with no measurements to back it up. What I'd have to do to actually find out is attach a temp sensor to a big sheet of aluminum and pump it. Even the small mass of an AD590 sensor is enough to prevent much of the cooling I am expecting is going on, here. I may actually mock this up sometime when we are not working on serious stuff here.

Jon

Reply to
Jon Elson

Fascinating.

formatting link

Michael

Reply to
mrdarrett

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.