Electrons Sloshing About: Resonant Versus Nonresonant SFG

Good afternoon, Darien!

Today, I would like to just briefly nerd out concerning a spectroscopy technique that is tremendously cool.  I have spent a lot of time over the past month shadowing Shuichi (a postdoc in my group) in order to learn how to operate his laser setup.  I find myself continuously asking “What is this optic?” or “Why did you move that optic?”, and the answers I get are usually pretty straightforward “That is a mirror” and “I moved it so that the beam would pass through this iris” and (increasingly) “I have told you this several times before.  Perhaps you should take notes this time”.  A while back, though, I got an answer that was really neat, and that is what I would like to share with you in this post. Before I dive into the glorious details, I should lay a bit of a background.

The setup we were working on was our sum frequency generation (SFG for short) setup.  Just in case somebody besides you reads this, I suppose that I should elaborate.  Sum frequency generation is a process in which photons of two different colors of light (red light and infrared light in my case) combine their energies to create a new photon with an energy which is the sum of the energies of the two initial photons.  This is a fascinating thing, because photons ordinarily don’t interact with each other at all.  You can point two flashlights so that their beams intersect, but the beams will continue right on without any combination, collision, or other observable change.  In order to make photons combine with each other, you need two things: VERY high density of photons, and a crystal or reflective surface to shine the light on.  Thus, like any other SFG setup, ours consists of a visible beam and an IR beam which both strike a surface (gold or platinum) at exactly the same position at exactly the same instant.1  After this, any SFG photons that are created go to our detector, and we see a nice spectrum of intensity versus photon frequency, with the center frequency lying at the sum of the frequencies of the red and IR beams we used (by beams, I of course mean pulses).

SFG works by a two step process.  First, the electromagnetic field of the IR beam sets up a sort of vibration of the electrons in the material being used, and then the visible beam strikes the surface.  The visible beam picks up energy from the vibrating field of the electrons that are sloshing about2 in response to the IR beam, which we observe as an increase in the frequency of some of the photons in the visible beam.3 Now, in practice, this doesn’t look like a two step process most of the time.  The electrons can dump their energy into the visible beam at pretty much the same instant that they are made to slosh about by the IR beam.  In fact, the electrons in a metal will lose all of their sloshy vibrational energy within something like 100fs, so if you are trying to get an SFG signal from a metal surface, you must make sure that your visible beam arrives at pretty much the same time as your IR beam.

We use a gold mirror for our SFG alignment, and we carefully adjust a delay stage while watching the signal from our detector.  When we see that the SFG signal is maximized, we are happy, because we have found “delay zero”, that marvelous delay stage position that makes the visible and IR beams arrive at the same instant in time.  But measuring SFG from gold surfaces isn’t our ultimate objective.  We want to measure the SFG signal from a carbon monoxide (CO) layer adsorbed onto a platinum electrode.  It turns out that carbon monoxide really likes platinum (just as it really likes hemoglobin), so it is pretty easy to get it to stick to the surface in a nice layer.  It also turns out that when CO absorbs IR light, the electrons in its molecular orbitals vibrate quite nicely.  Unlike the sloshy vibrations in a metal, though, these are nice regular oscillations, that are said to be “resonant”.  To distinguish between this sort of excitation and that in metals, we call processes “non-resonant” if they involve the latter type of excitation.

Now many of our experiments thus far have been using CO as a probe to learn about the conditions at the surface of an electrode when a potential is applied, so after using a gold mirror to find delay zero, we exchange it for a platinum mirror covered with an adsorbed layer of CO.  Once this is in place, we can look at our signal and see a really big SFG peak.  This makes us happy, because everybody likes a high signal to noise ratio.4 To celebrate, we do something very strange: we go back to the delay stage and we move it to delay the arrival of the visible beam until hundreds of femtoseconds AFTER the IR pulse.  This makes our glorious SFG signal go down! Our signal to noise ratio is still good, but it has taken an undeniable hit.  When I first saw Shuichi do this, I was rather distraught and demanded an explanation.  “Why did you move the delay stage???” I asked.  “You decreased the signal!!”.  And then Shuichi proceeded to give me the fantabulously cool explanation that I am about to share with you.5

As you might expect, not all of the IR light is absorbed by the CO layer.  Lots of it passes right through and excites the electrons in the metal, making them slosh about.  Thus, if the visible beam is made to arrive at the same time as the IR beam, the SFG signal that it generates will be a combination of the nonresonant signal due to the electrons in the metal, and the resonant signal due to the electrons in the CO molecular orbitals.  This is not good! Our data is hard enough to interpret already.  I don’t know what we would do if we had to account for some unknown mixture of signals from the CO and the platinum surface beneath it.  Happily, everything is resolved by delaying the visible beam.  Although the nonresonant excitation of the electrons in the metal is very short lived (as already stated), the resonant excitation of the electrons in the CO molecular orbitals is much longer lived.  Thus, if we delay the visible beam by a few hundred femtoseconds, the nonresonant excitation will be gone before the visible pulse arrives, but the resonant vibration of the CO bonds will be almost as strong as ever.  In this scenario, the only SFG signal that can possibly result is the resonant SFG signal from the CO!  We may sacrifice a little bit of our resonant SFG signal, since some of the CO vibrations will decay slightly while we wait for the nonresonant excitation to dissipate completely, but it’s totes worth it!

Have a marvelous afternoon!


  1. It turns out that it isn’t always necessary to arrive at exactly the same instant, and this will be elaborated on in a couple of paragraphs.
  2. This phrasing was used by Dana when discussing these things with me.  I like it so much that I will proceed to use it repeatedly throughout this post.
  3. I am told that the interaction with the IR is a “polarization” sort of interaction, while the second step is a “Raman” sort of interaction, but I didn’t understand fully enough to elucidate that in more detail.
  4. No citation is needed for this statement, as it is common knowledge.
  5. The next paragraph is what I really wanted to share.  It may sound a bit anticlimactic, though, because I was compelled to spread much of the coolness of it throughout the paragraphs of background information.

One thought on “Electrons Sloshing About: Resonant Versus Nonresonant SFG

  1. This is indeed uber-cool! I really like your explanation of SFG—I normally think of this things by their proper names like “oscillating polarization” or “coherence” instead of the picturesque and intuitive “sloshing about”. I hadn’t ever thought to use the coherence lifetimes (how long it takes the sloshing to stop) of the metal vs. the CO molecule to my advantage. But it is a great usage!

    In my group we use non-resonant driven signal (what you are talking about in the metal) to find zero delay for three laser pulses. We do this in CCl4 (if I remember correctly).

    We have to deal with this non-resonant background signal all the time in our experiments. It is a big pain. There are just so many more substrate oscillators compared to sample oscillators, that sometimes all the signal we get is from this non-resonant material. These are sad days, especially when we initially thought we got signal on a super cool material only to find out it was just the sapphire under the material.



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s