Amid all the terrible news that’s been going around lately, I figured I’d post something nice.

A couple of weeks ago we submitted another manuscript and it’s currently with referees. In it we discuss our observation of step-flow motion occurring on a platinum surface at very high temperatures. We actually do it with a new trick as well (and that’s actually the most important part). I’ll try to explain it below in terms that I hope are understandable.

Imagine a perfect surface and we’ll shine a beam of light (x-rays in this case) on that surface. Let’s say something like this, with a diagram of the surface we’re scattering from (on the left) and a calculation of what would then be observed in our detector (on the right):



The illuminated area has no structure, therefore the scattering we observe is a single bright region with a few fringes, all related to the size and shape of the incident light (I’ll ask that you forgive that the simulation is from a square beam and not the circular one shown).

Now, we’re going to add a couple of complications. First, we’re going to look at one of those edges instead the middle of a terrace. Next, realize that there is a small addition distance if you reflect from the lower side than from the upper side. Now, remember that we’re using x-rays. And what I’ve also not told you is that we’ve got the angles of everything set such that the extra distance adds up to 1/2 the wavelength of the photons. If it’s exactly 1/2, then the two regions can cancel each other and you see something like this.



The single spot of scattered light splits into 2 regions, with zero in the middle.

Recall that I said the platinum was hot. It’s not just hot, it’s really hot, usually around 1800 K. At these temperatures the surface will slowly sublimate (go directly from solid to vapor). Very often atoms at the step edges (particularly at kinks in the step edge) and diffuse around on the surface. At these temperatures the atoms will often then leave the surface through sublimation and never return. What you find is a net loss of atoms occurring from the step edges. As the atoms leave, the position of the step shifts (in this case to the right).

I suppose the only other piece of the puzzle missing (and it’s an important piece) is that the beam of photons we’re using is highly coherent over the illuminated area. Practically what that means is that a single photon can interfere with itself with contributions coming from the entire (almost) illuminated area. That’s important. If the “coherence” was smaller (as is usually the case in scattering experiments), then you’d have lots of photons that only see the top step or the bottom step even when the step-edge runs through the middle of the illuminated area.

So, we’ve got steps traveling from left to right and we’re sensitive to changes over the entire illuminated area. Occasionally there is a step in view, occasionally there is no step. When the average distribution of the steps in uniform (think of a staircase, though with very small vertical steps!), then what you’d expect is that the observed scattering pattern oscillate between the two patterns shown above. And not just oscillate, but do so in a rather uniform fashion. This is exactly what we find!

On the left we have the “perfect” surface with a single spot more or less in the image. On the right is an example for a few tens of seconds later where single peak has split and is now 2 peaks. If we continue to wait, the image will eventually resemble the first image again. This cycle repeats for many minutes before becoming noisy again (ie more structure and dynamics than just single steps moving).

We can then analyze (it’s a kind of averaging mostly) the images to produce a nice signal. I’ll show you 2 examples, though we’ve measured 10 or so.

These are a couple of the signals we are capable of detecting (I’ll spare you the details of how we get to this point). The important thing is that you can see a fairly clear signal, an oscillation. The period of oscillation (how long it takes to go from a peak to another peak) corresponds to the time it takes for a single step to move across our field of view. What can you get from this? Well, first, at higher temperatures the oscillation is faster. That makes sense as the higher the temperature then the faster atoms should leave the surface and hence the faster the steps will move. Also notice that the higher temperature signal appears “more ordered” than the one at lower temperature. This indicates that as the temperature increases step-flow motion becomes more important than other motion (such as step-meandering). There’s a third and fourth thing in this signal that take us in a different direction and relate to something we’re working on for a different paper (they’re related to a more general behavior of the surface even when it has several steps/defects).

“Ok,” you say, “That’s nice, but how can you be sure what you’re seeing is real?” Great question. The rate of change of the oscillation frequency with temperature can tell us how much energy it takes for atoms to sublimate. In this case we measure 5.4 (.9) eV. That can be compared to the known value of 5.9 eV. So, to within our ability to measure, the temperature dependence of the frequency ties exactly to a known quantity. That’s a pretty strong indication that we’re doing things correctly. (I’ll spare you all the details of how we repeated the experiments twice with difference samples/configurations to help convince ourselves)

I’m pretty sure this is the first observation of step-flow motion on a platinum surface. But more importantly, this is a different way of measuring such processes than through the usual means of electron microscopy. Since this is an x-ray based technique, we can do such studies in environments that are not just vacuum measurements (like this). Instead we can measure in harsh chemical environments where electron microscopy has difficulty. And to give a hint, we’ve already been doing this! Additionally this serves as a nice connection between x-ray photon correlation spectroscopy (which measures dynamics) and imaging structure.

In case anyone is really interested, a pre-print of the paper is available on the arXiv servers, cond-mat > arXiv:1103.0263 Persistent Oscillations of X-ray Speckles: Pt (001) Step Flow. I’ll be presenting this along with some newer work next week at the American Physical Society March Meeting in Dallas.