Since DCG is orders of magnitude more sensitive to 445 nm than to 532 nm, they would be ideally suited for holography provided it was possible to get them to lase in a single mode. At first, prospects were dim because the lasers were believed to consist of multiple emitters. To get a single mode output, not only would it be necessary for each emitter to lase on a single mode, but on top of that, all emitters would have to lock onto the same wavelength. They tend to do this naturally however, because there is some coupling between the emitters. Because of this coupling, the gain is slightly higher for oscillation modes of the whole assembly ("supermodes") that contain little power in the spaces between the emitters. So the pattern that's most likely to form, is also the least desirable: a 180 degrees phase shift between the odd and even emittters in the array. This means that because of destructive interference, power is emitted in all kinds of directions except the most important one: the forward direction. The situation is a bit like having a 100% efficient grating right behind the facet of a normal laser diode completely removing the zeroth order.
Soon however, it became clear that either the diodes must be just broad single emitters or the locking does end up favorable in these diodes because the first reports of single mode operation came: http://hololaser.wordpress.com/
Of course at that point (or somewhat earlier to be honest) I couldn't resist the urge anymore to buy a projector and (while chanting and praying for the gods of the Sacredness of Untouched Electronic Equipment to forgive me) extract the diodes from it.
To investigate the spectrum this time, I attached a webcam with the lens removed (sorry again gods :-) ) to a dual grating monochromator, the both of them together acting as a high resolution spectrograph. This setup has a higher resolution and is more sturdy than the setup I used before to test the mode structure of other lasers ( http://piepklein.blogspot.com/2009/01/modes-and-laser-diodes.html )
To get a general impression of the mode structure I ramped up the diode current in steps:
You can see the typical evolution in the mode spectrum of a high power laser diode. Near threshold, the emission is single mode or 'couple of modes'. As the current is increased, more gain in the cavity allows more modes to lase simultaneously.
As the current is increased further however, the mode spectrum for this diode seems to split into two parts. At the right is a part that looks a bit fuzzy. Even though the optical quality of the monochromator system is not excellent since it's not designed for spectral imaging, I do believe the fuzziness is real, not an artifact. But more about that later in a follow up post.
Walking away to the left is the other part. Once in a while a strong mode pops up, but as the current is increased, this part loses it significance quite quickly.
So, how to get the multimode spectrum tamed to show a nice clean single mode? The answer is to provide external feedback. If you can provide some extra gain for one mode and/or suppress the gain of the others, you can force the diode to lase on that single mode.
The feedback mechanism should have enough wavelength selectivity to single out the desired mode. This selectivity can be obtained by using the properties of the extra fabry-perot cavity that's formed between output coupler and a third mirror that's placed outside the cavity and by using a grating to reflect only a small band of wavelengths back into the cavity.
The increased feedback will also cause the threshold current of the laser to decrease. Because less light escapes the laser at the output coupler the net round trip gain to sustain lasing does not have to be as high as without extra feedback. Since the gain scales with the current through the device, the current can also be lower and thus the laser will lase already at lower currents.
To test how easy it was to influence the modes I had already tried to see how much I could influence the treshold current (this was before the spectrograph was finished). With the current set just below threshold and using a lens to focus the collimated beam to a small focus I put a glass microscope slide at this point. I was surprised to see that indeed the laser started to lase again and after fiddling a lot I was able to reduce the threshold current from 218 to 132 mA with just the 4% feedback from the glass! The laser was indeed very susceptible to feedback!
So, what does the feedback do to the mode structure? In the next clip I'm ramping the current up again in the lower range. Instead of the reflecting the light back by putting the glass plate in the focus of a lens, this time I just put the glass plate directly in front of the laser diode, without even collimating the beam first. The level of feedback must be very low in this setup. Yet look what happens:
Between around 220 and 300mA the laser now operates on a single mode almost constantly (the multiple modes right after each step are probably due to the die reaching a new thermal equilibrium).
To check if it's really the feedback from the slide that's causing this, in the next video I'm deliberately moving the slide towards and away from the diode, by gently tapping on it. Because this shifts the modes of the external OC+external mirror cavity, you can see the wavelengths at which the whole assembly lases sweep along. Because the finesse of the external cavity is very low, the laser is only single mode on the upper and lower extremes of the gain profile when the external feedback pulls it towards an edge. But with some collimation and/or a higher reflectance mirror, this can probably be improved.
Even though a good stable setup would require a TEC and a piezo to position the slide at the right distance or even a true ECDL, I feel time's almost right to soak some gelatin, make some plates and let the IPA-fumes fill the air again :-)