Over at Uncertain principles, Chad talks about how Nobody Expects Bose-Einstein Condensation, i.e. while the phenomenon had been predicted, the enabling technology was serendipitous.
What really made [magneto-optic trapping] take off, though, was that people figured out you could get the laser cooling wavelength for rubidium from diode lasers. And diode lasers are manufactured in mass quantities because they’re used in CD players, laser printers, and other commercial electronic devices. So, rather than needing to spend a couple hundred thousand dollars to get a dye laser system up and running, you could get a working laser system for a couple of grand.
And it’s true. The main enabler was the availability of diode lasers. And their ability to be tuned electronically and thermally. Ah, two! The the two main enablers were their availability and their ability to be tuned, and their susceptibility to optical feedback. Oh, three! The three main enablers were their availability and their ability to be tuned and their susceptibility to optical feedback. Hmmm. Among their advantages are such diverse elements as their availability and their ability to be tuned and their susceptibility to optical feedback, and a nice red color. Damn.
The main enabler was their availability. Blah, blah,blah.
OK, I’m kidding a little, because without the technical advantages, who cares if it’s available? But it’s an excuse to do a Python bit, and talk about the other things.
Laser diodes are pretty neat, though somewhat fragile. In the days of building my own systems, I never knew one that died a natural death-from-old-age. They all got blown up somehow, and I wish that had been in the days of digital cameras, because I recall one that looked cool under the microscope, with one facet blown off but still the piece still hanging there, attached to a tiny wire that was part of the circuit. You had to remove the can that surrounded the diode before you could mount them in our homemade system, and that took some practice. But there were usually some dead or otherwise useless diodes around for practice for newcomers to the lab.
Being really high gain meant only a few passes of light for the lasing medium — the output facets reflected only about 35% of the light, and for higher-power diodes the usable facet was antireflection-coated. This makes them very susceptible to optical feedback; light sent in would “tell” the diode where to lase, so the strategy was to reflect light off of a grating, and tip the grating to tune the wavelength of light being fed back in. Coarse tuning was provided by the temperature and current; usually you stabilized the temperature once you were close, and adjusted the current tuned the grating angle to get you to the target wavelength. You could lock to a spectroscopic line using feedback to the grating and/or to the current, depending on the feedback bandwidth.
Rubidium was the workhorse, and there are two isotopes (Rb-85 and Rb-87) and I’ve trapped both. Laser diodes also allow trapping of Cs-133, and I’ve done that as well. I’ve also trapped K-37, K-38m, K-40 and K-41 using a Ti::sapph pumped by an Ar-ion laser, back in my postdoc days. The former of those two are radioactive (well, technically so are K-40 and Rb-87, but with a half-life of a billion years or so the decay does not present much of a limitation). The first two, however, have half-lives of around a second, so they tend to decay in your trap, which is ideal if you want to do experiments that rely on their decay. Other folks trap radioactives, but of a much longer half-lives. My position has been that if the trap lifetime is shorter than the decay lifetime, then the atom is effectively stable. (That’s just me, though, in response to other trappers bragging on how many atoms they had trapped.)