Watching Ytterbium

Absorption image of Yb ion.
Image source. Via.

If you know anything about atoms you know they're small. And if you know a little more you know that the typical size of an atom sounds Swedish - it's a few Ångström, or 10^-10 meters.

First actual images of atoms went around the world two decades or so ago, taken with scanning tunnel microscopes. These microscope images require careful preparation of the sample, and also take time. It is highly desirable to find a method that works faster and is more flexible for small samples, ideally without a lot of preparation and without damaging the sample.

Taking an image with a scanning tunnel microscope doesn't have a lot in common with watching something the way that we are used to. For the average person "watching" means detecting photons that have been scattered off objects. Quantum mechanics sets a limit to how well you can "watch" an atom absorbing and releasing photons of some energy. That's because the absorption of a photon will excite an electron and temporarily put it into a level with higher energy. Alas, these excited levels have some lifetime and don't decay instantaneously. As long as the electron is in the excited state it can't absorb another photon.

So you might conclude it's hopeless trying to watch a single atom. But a group of experimentalists from Australia have found a nifty way to do exactly that. Their paper was published in Nature two weeks ago

Absorption imaging of a single atom
Erik W. Streed, Andreas Jechow, Benjamin G. Norton and David Kielpinski
Nature Communications 3, 933, (2012)
arXiv:1201.5280

So how do you do it? First, get some Ytterbium. Strip off an electron, so you have a positively charged ion, and put it into an ion trap in ultra high vacuum. Then laser cool your ion to a few mK (that's really, really cold).

Ytterbium has a resonance at 370 nm (in the near ultraviolet). At that frequency you can excite an Yb electron from the S ground-state to the P excited state. Alas, if it decays, the electron has a probability of 1/200 to not go back into the ground state, but end up in a metastable D state of intermediate energy. The lifetime of the excited P state is some nanoseconds, but that of the metastable state is much much longer, about 50 microseconds. So if you just keep exciting your atom at 370 nm, after some nanoseconds you'll have kicked it into the metastable state where it stays and you can't watch anything anymore at that frequency. So what's the experimentalist to do? They stimulate the emission with the right wavelength, in this case at 935.2 nm (in the near infrared), to get the electron back from the metastable state into the ground state.

Actually, to excite the atom you don't need incident light of exactly the right frequency, and in fact that's not what they use. The absorption probability has a finite width and is not exactly peaked. That means there's a small probability the atom will absorb light of slightly smaller frequency and then emit it at the resonance frequency. The actual light the experimentalists used is thus not at 370 nm, but at 369.5 nm. That has the merit that you can in principle tell (with a certain probability) which light was absorbed and reemitted and which one was never absorbed to begin with. The detuning also gives you a handle on how strongly you can afford to disturb your atom, for every time a photon scatters off it, it gets a recoil and moves. You don't want it too move too much, otherwise you'll get a blurry image.

So here's then how you take your image. Shine the slightly detuned light on the ion while driving the transition back from the metastable state to the ground state, and measure the photons at the resonance frequency. Do the same thing without driving the transition back from the metastable state. This has the effect that the probability that the ion can absorb anything is really small and you get essentially a background image. Then subtract both images, and voila. While you do that, you better try not to have too much fluctuations in the intensity of the light.

The merit of this method is its flexibility and it's also reasonably fast with illumination times between 0.05 and 1 second. The authors write that with more improvement this method might be useful to study the dynamics of nucleic acids.