I ended the previous laser post by noting that diode lasers need some additional wavelength selection to be done in order to be useful as light sources for spectroscopy experiments. In their natural state, they tend to emit light over a broader range of wavelengths than is really ideal, and we’d like to narrow that down, and also to be able to control the emission.
(I should note that, while the emission of a typical diode laser is broader than people doing atomic physics experiments would like, it’s still incredibly narrow by normal standards. The actual width, in wavelength, of the light emitted by a free-running diode laser is something on the short side of 0.003 nm for the 811 nm lasers I’m interested in. That’s pretty good in an absolute sense, but it’s a factor of 10,000 or so too big for my applications.)
There are a bunch of techniques that people use to narrow and tune diode lasers. I’ll only discuss the two that I use in my lab, after the cut.
The easiest way to get a given diode laser to the wavelength you want is to have another laser that’s already running at the right wavelength, and stick some of that light into the cavity of the laser you want to tune. This is referred to as “injection locking,” because you inject some of one laser into the cavity of the other, whereupon the two wavelengths become locked together.
The basic idea here is to take advantage of the stimulated emission process. If you just wait for things to happen naturally, you’ll get spontaneous emission at a bunch of different wavelengths, and each of those photons will trigger stimulated emission, and create more photons at whatever particular wavelength they happen to be at, giving you broad emission. But, if you stick in a whole bunch of photons at a particular wavelength, you can sort of drown out the random noise caused by spontaneous emission, as the photons you’re sticking in will outnumber the spontaneous photons, and account for the vast majority of the stimulated emission. Which, in turn, creates even more photons of that wavelength, and more stimulated emission, and so on.
Of course, at this point, you might sensibly point out that if you already have a laser with the right properties, there’s no reason to bother. The reason to do this is that it can basically act as a light amplifier– if you have a cheap, low-power laser at the right wavelength, you can use a little of its light to inject a more powerful, more expensive laser (or two), and get a lot more light out. In my lab, I use something like 2 miliwatts of light from one laser to injection lock a second laser that puts out about 150 mW, which isn’t too shabby.
Of course, there’s got to be a way to make the low-power laser do what you want in the first place. The technique here is sort of a cross between injection locking and the techniques used for dye lasers and solid-state lasers. In a dye laser, you put a wavelength-selective element inside the laser cavity, and that picks out a narrower range of wavelengths to be amplified. You can’t do that with a diode laser, as the cavity is a hundred-nanometer layer of gallium arsenide, but you can put a wavelength-selective element on the outside of the cavity, and use it to send some light back in.
The element of choice is a diffraction grating, which is basically a mirror with a very regular pattern of very small grooves cut into it– something like 2400 grooves per millimeter. In addition to the normal reflection you expect from a mirror, this also causes diffraction, with different colors of light being spread out to different angles. It’s the same effect caused by the tracks on a CD, that gives you that spiffy irridescent effect, only the grooves are much closer together.
If you align the grating just the right way, you can arrange it so that the light that is reflected in the normal way goes out to the rest of the apparatus, while some of the diffracted light is sent back the way it came, back into the laser cavity. And since the diffraction angle depends on the wavelength, by changing the tilt of the grating, you can change exactly what wavelength gets sent back into the cavity. You can sort of think of this diffracted light as “injection locking” the laser to a narrower range of its own emitted light– if the grating is properly aligned, it will select a very narrow wavelength range to be amplified, and the light that is reflected out to the rest of the apparatus will have the desired characteristics. You can tune the exact wavelength by changing the angle of the grating, to select whatever specific wavelength you’re after– the diode system I use can be tuned over several nanometers using the grating.
This is a very effective and very common technique, and serves as the basis for commercial tunable diode systems. It’s also relatively easy to build your own– the lasers I have are in mounts that were machined and assembled by undergraduate students.
There are, of course, a number of disadvantages to the technique. For one thing, you’re throwing away a good chunk of the laser power to do the locking. Also, the alignment is kind of fiddly, and even with good mirror mounts, the laser will drift out of alignment over time. The tuning also introduces problems for optical alignment, as changing the grating angle also changes the direction of the output beam, meaning you have to re-align everything downstream. And, of course, if you have a situation where the conditions in the lab change really dramatically in a short time– say, the humidity goes from 30% to 80% in the space of about two days– then you can get some crap on the surface of the grating, which completely wrecks it, and requires the entire thing to be rebuilt. Just, you know, hypothetically speaking.
Anyway, those are the basics of diode laser operation and tuning. In my lab, I’m using a homemade grating-locked diode system, shown here (click for a larger image). The laser itself is mounted inside a temperature-controlled aluminum block (the big grey thing mounted on a copper block, which serves as a heat sink), while the grating is attached to a mirror mount (the black thing at left). The output beam reflects off the grating, heading up and to the right, where it hits the two mirrors in the upper right of the picture. The thing that looks like a pen is a pen, for scale.
This laser gives me something like 20 mW of total power, a good chunk of which is used for various diagnostics and frequency stabilization. The rest of that light is coupled into an optical fiber, and used to injection lock a 150 mW diode laser on another part of the table. That 150 mW will be sufficient for all of my laser cooling needs, once I finish repairing the damn grating.
And now you know what I’m working on at the moment.