Today in the lab, I saw something very odd.

We have two vacuum chambers, Main and Load-Lock, each with a vacuum gauge. They are separated by a gate valve (a vacuum tight sliding door). There is a pump emptying out the Load Lock (the Main also has a pump, but it's closed off and out of the picture right now)

To begin with, Main chamber is at 45 µT.
Load-Lock is initially at 15µT and is dropping only slowly as I'm pumping on it (>15 sec to get from 16 µT to 15µT).

I open the gate valve. As expected, the pressure in Main drops rapidly, and for a moment the pressure in Load-Lock rises a bit. However, within 15 seconds, both are at 5 µT.

That is, after I opened the connection between the two chambers, the pressure in both chambers dropped in pressure, far more rapidly than the small chamber was pumping down to begin with.

We have three ideas on how that could happen. What are your thoughts? What would you check next?

First i would check, what happens if the gate is closed again. I have no idea, how the gate is constructed, but propably the opening the gate somehow makes more volume for the vacuum.

Second i would check other moveable parts, if they could lead to more volume and propably could have been moved due to preasure differences. The pump and the disabled pump could be primary suspects.

Not knowing how exactly the pump works, could it be, that the vacuum is actually created in some chamber that is part of the pump, and the exchange with the load lock has not worked properly.

I also would check the preasure measuring instruments in the load lock. It might be, that some mechanical part got stuck and it gave too high values. Opening the gate and the resulting preasure changes did shake it loose.

Would there be any reason, for the main chamber to be much colder then the load lock?

I agree with arcosh about the gauge being stuck. That would also explain why the pressure was dropping so slowly (unless that's a normal rate for your pump.) Have you tried re-creating this, possibly with new gauges?

If we close the gate valve, the pressures remain low. Everything seems to be normal now.

The vacuum gauges are ion gauges - no moving parts.

The main chamber pump is a cold trap, and is cut off from the chamber by a gate valve. The main chamber is room temperature.

The precise operation of the pump isn't relevant - it's a turbopump (basically a very very steep many-bladed fan spinning ridiculously fast), and so can be counted on to actually provide a substantial pressure difference regardless of what else is going on. The valve connecting it to the load lock was wide open and functioning fine - it wasn't sucking the air out through a straw when the load lock was near atmospheric pressure.

How do the gauges work? How are the results of the gauges displayed? Maybe you just have a Gauge-GUI problem.

Is there anything in the chambers?

The gauges work by measuring an ion current flowing between two very differently charged wires, which are kept hot. This ion current will be proportional to the number of air molecules. The gauges' results were displayed on LCD screens.

The chambers have things, but nowhere near the action. Remember, what caused the drop in pressure was a gate valve opening. A sliding door sliding open.

Could there have been some dirt between the wires of the gauges, that was blown away, when the air rushed in? Or if the wires are under some protective cover, air exchange between the chamber and the area between the wires was blocked.

I suppose that there is no chance, that the 2 chambers had originally different gases, that are differently conductable?

Could there have been some weired air current, that concentrated air at the gauge? That was changed, due tot he door being open. If so, closing the door and starting the pump again, would raise the preassue again and switching the pump off make it drop again. I could not come up with a design that would lead to such currents though.

An other straw, that i could reach for, would be some chemical reaction, that did bind some part of the air.

Assuming the gauge did work correct, either the volume of the vacuum had to increase, but we have ruled out pretty much everything that is part of the chambers themself. If anything inside the chambers decreased it's volume that could work too, but i suppose that would have had a visible effect, that you would have seen already.

Or some of the air had been removed. Though i can't come up with anything that would remove air, especially not something that would be tiggered by the opening of the gate. It would have to be something in the main chamber, that happens either on a sudden preassure drop or at a certain low preassure, or in the load lock that is triggered by a sudden increase in preassure.

Or it could be a temperature drop (though i don't know, by how much, to get the preassure differences you noticed).

Overall i would consider a gauge failure most likely, since it can explain both the sudden (appearent?) not working of the pump, and the sudden preassure drop later.

Our theories:
1) it was a 'virtual leak' in the gate valve. That is, there was a tiny pocket of air trapped inside the gate valve (behind it, in the slot in the wall it slides into). As this air slowly bled out, it degraded the vacuum in the load lock. By opening the gate valve altogether, the air was suddenly released, which finally allowed it to be pumped away completely. The pressure spike, though large, was over before the pressure gauge could respond, so we only really saw the decrease.

Because we'd been pumping on the chamber for a while, getting it down as far as 15 µT the slow way, the walls were almost entirely outgassed (i.e. the water and other quasi-volatile deposits on the walls had evaporated), so rapid pressure drops upon removing a leak are possible.

2) The gate valve had something stuck on it that was gradually evaporating. By closing the gate valve, it was scraped off, which radically reduced its surface area, and thus the rate at which it was evaporating. Or perhaps it was left behind in the gate valve housing when the valve was re-closed. Or perhaps it finished outgassing while the valve was open, before we re-closed.

I don't remember our third possible explanation.

Yes thoose make sense too. After my last post i had actually thought about something evaporating, but i did not come up with an idea, how that could be connected to the gate.