Quantum Optics VII: Sky lasers

People are interested in measuring the varying chemical content of the atmosphere for many reasons. We’re interested in pollution emitted from power plants and the potential release of toxic gases into the atmosphere by terrorists, for example. These situations often involve very small amounts of the undesirable chemicals and are difficult to detect. If we’re worried about the release of toxic gases, we don’t want to send an airplane through the cloud of interest to collect information on what’s in it, and we want to know long before the cloud drifts toward us that we’re dealing with something dangerous.


In my Raman spectroscopy blog, I talked about our lab’s success at detecting trace quantities of anthrax using a variation of Raman spectroscopy. This inspired the idea of using Raman to detect atmospheric pollutants, since the sensitivity of Raman is unmatched when it comes to detecting tiny amounts of chemicals.


There are a few problems, though. Most obviously, clouds are really far away. Beyond that, even if we could shine a really high powered, well collimated laser up into the cloud, only a fraction of the laser photons will make it to the desired destination. On top of that, the scattered Raman signal is so weak, there’s no way enough photons would make it back to our detector on the ground to tell us what we’re looking at. They would get absorbed, scattered, or travel out of the solid angle of our detector. So you get loss both ways. No good.


What if we could put the laser on the other side of the experiment? Put the laser up inside the cloud and collect the photons on the ground to see what travels toward the detector (this is another spectroscopic technique; sometimes you look at what comes back at you, sometimes you look at what doesn't come back). Then we’d only have to fight loss in one direction. Better, but that loss is still too significant for us to detect just a few molecules. What we really need is some way to enhance this signal as it travels through the atmosphere.


This sounds like something that just isn't possible, but as you have gathered from the previous topics we've touched, quantum physics allows us to do a lot of seemingly impossible things.
Before I get into the idea of the “sky laser”, let me cover my references. This work is the brilliant idea of the professor I work for and several members of his group, and I’m going to keep my explanation here limited to the information included in the following paper (check it out if you want the details, it isn't too hard to follow):
http://acms.arizona.edu/MURI/Publications/37_standoff.pdf

So now on to the physics...

The goal: detect a few molecules of some toxin in a cloud somewhere in the sky.
How we propose to do it:

Getting a Raman signal is easy enough; shine a laser at a sample, it absorbs light, emits light, and you collect the emitted light and identify the molecule (see the Raman post linked above). The issue is getting this signal to be large enough, by the time it travels from the cloud back to the ground, to be detectable.

Aside:
+ Show Spoiler +


Okay, pictures speak louder than words, so here we go:

+ Show Spoiler +

We first send out two femtosecond pulses separated by a time ∆t1,2. They have slightly different wavelengths and therefore travel at slightly different group velocities; we pick the wavelengths such that the second pulse travels faster than the first. Then, the second pulse will eventually catch up to the first and they will overlap; we choose ∆t1,2 so they overlap toward the back of the cloud of interest. These pulses excite the molecules of interest into a particular state, and in turn they will de-excite by emitting a photon corresponding to the excitation energy.

This light will be emitted in all directions, since at this point it is mostly spontaneous emission occurring (we just drew one dimension in the image). We care about the stuff emitted in the backward direction, because we’re sending out another set of pulses separated by a smaller ∆t3,4; these will overlap a little to the left of the first set of pulses.
So we have a group of excited atoms right next to this group of spontaneously emitted photons that have the exact energy needed to stimulate emission in these atoms. (sound familiar?) And stimulate emission they do! These emitted photons are going to travel in the same direction as the photons that stimulated the process: back toward the detector on the ground!

We keep sending these sets of 2 pulses and keep decreasing the value of ∆t so that we’re constantly preparing increasingly closer sections of the cloud for stimulated emission. This produces what they call “swept gain”; we’re getting continuous gain but instead of using a cavity like in a laser, we’re preparing the gain medium (the atmosphere) by sweeping the preparation pulses across the medium.
The result is that this light emitted from the particular molecule of interest travels back toward the detector on the ground, and the signal gets increasingly larger as it travels. This allows us to detect very small quantities of toxins because we only need a tiny signal to get this stimulated emission process started. We were able to do this by sending out laser pulses from the ground, too. We didn't have to put the laser on a plane and risk the pilot’s health to do the experiment.


Unsurprisingly this is very difficult to carry out experimentally. A femtosecond is really tiny, and those are the time scales we’re working with. Selecting the ∆t’s, the wavelengths, knowing very well the dispersion of air as a function of position in the atmosphere… all of these things are really difficult and have to be controlled to a very high accuracy or we won’t get the gain we want. So rather than going outside and trying these experiments in the real atmosphere, our researchers have been working with pure oxygen and pure nitrogen in the lab, and are simply trying to generate this amplified backward propagating pulse. They’re having some luck, too. It turns out that this is actually probably going to be demonstrated successfully at some point within the next few years, so that’s kind of exciting. Once we get a better idea of what is possible in an “idealized” situation, we can begin tweaking the technology to make the method more robust and suitable for mainstream implementation.