Every Picture Tells a Story

So, what's the deal with that graph I was talking about the other day? I sort of left it hanging at the end, there, but I ought to actually interpret the figure. It also serves as a nice and fairly simple illustration of how physicists approach experimental data.

Here's a newer version of the plot in question, made using more recent data:

shows the fluorescence signal detected in our metastable atoms source as a function of the gas pressure on the source inlet. There's a clear peak in the data at a pressure of 60-70 milliTorr (ignore that one point flying way up above all the others-- that's a fluke), dropping rapidly to zero at lower pressue, and tailing off a little more slowly at higher pressure.

Both of these effects make perfect sense, if you think carefully about what we're looking at:

The low pressure side is easy to understand. At low pressure, the signal is very small simply because there aren't any atoms there to be excited. The pressure is, loosely speaking, a measure of how many atoms you can expect to find in the cubic centimeter or so where the atoms interact with the laser. If there aren't any atoms, there can't be any fluorescence, so we would expect the signal to increase sharply as we increase the amount of krypton in the system.

A naive guess at what this should look like would suggest that the signal should increase linearly at low pressure-- that is, if you doubly the amount of gas, you should double the amount of fluorescence. If you squint a bit, you can convince yourself that you could draw a straight line through the data up to about 40 mT, at which point some other effect takes over.

The tail-off at high pressure is a little more complicated, and probably involves at least two different effects. One of them has to do with the nature of the atoms we're looking for: we can only detect atoms that are in a metastable state, with a significant amount of internal energy above the ground state. Atoms excited into that state will stay there practically forever (about 30 seconds) before they decay naturally, but they can be knocked out of that state by collisions with other atoms. As the amount of gas in the chamber increases, the collision rate goes up dramatically (roughly as the square of the pressure), and eventually the metastables start to be destroyed faster than they're created.

The other effect has to do with the nature of the excitation process, which requires ground-state krypton atoms to absorb one ultraviolet photon from our really expensive lamp, and then absorb a second infrared photon from one of our lasers. If the atoms don't see both photons, they won't be excited, and we won't see any signal.

As the pressure increases, the amount of gas between the lamp and the interaction region also increases. The lamp is pretty close to the interaction region, but there are still a few centimeters of gas there, and when the pressure gets high enough, that gas becomes "optically thick," meaning that it absorbs most of the ultraviolet photons before they can get to the region where the infrared laser is.

If you put together an effect that causes the signal to increase with increasing pressure, provided the pressure is low, and an effect that causes the signal to decrease with increasing pressure when the pressure is higher, you get, well, a peak.You would expect a linear increase at low pressure, as you increase the number of atoms available to be excited, and a slower decrease at high pressure, as collisions and absorption reduce the signal, and that's just what we see.

Both collisions and absorption will (at least potentially) contribute to the drop-off of the signal as the pressure increases, and they both sort of do the same thing, in terms of the shape of the curve. One effect may be more important than the other, but it's a little hard to say which. Fortunately, thinking about it a bit does suggest an experiment to check.

We have a photodiode inside the chamber that is designed to detect light at the 123 nm wavelength of the ultraviolet photons used for the excitation. It's on the far side of the interaction region from the lamp, so all we need to do to test how significant the absorption effect is is to feed some krypton gas into the chamber, and monitor the intensity of the light reaching the diode as a function of the pressure. If there's a significant amount of absorption taking place, we should see the signal decrease as the pressure increases, and we can compare that decrease to what we see in the fluorescence signal.

Happily, this can be done without turning the lasers on-- we just need the lamp and the photodiode. This is a good thing, because it's unseasonably warm here in Schenectady. That means that the overtaxed air-handling system in our building has more or less given up trying to maintain a constant temperature in my lab, so it's six or seven degrees (Fahrenheit) warmer than it was a week ago when these data were taken. That's enough of a difference to throw all the lasers out of whack, and it'll take the better part of a day to get everything running happily again.

And that would be wasted effort, because it's supposed to get cold again over the weekend, so by Monday the temperature will be back down where it was last week, and we'd just have to re-tweak everything all over again. It's easier to just hold off on the laser tweaking until Monday, when we can expect an extended period of more or less constant temperature.

But, as I said, the simple experiment to look at the absorption effect doesn't require the laser, so that's what I'll be doing tomorrow. There's also a quick check that I can do involving some aspects of the background to our signal, but I'll hold off on talking about that until I actually know what's going on.