In 1985 Ashkin, Leo Hollberg, John E. Bjorkholm, Alex Cable and I at AT&T Bell Labs were able to cool sodium atoms to 240 millionths of a kelvin. Because the light field acts as a viscous force, we dubbed the combination of laser beams used to create the drag force "optical molasses." Although not a trap, the atoms were confined in the viscous medium for periods as long as 0.5 second, until eventually they would leak out of the cooling beams.
Optical molasses enabled us to solve the three major problems that stood in the way of constructing a laser trap. First, by cooling the atoms to extremely low temperatures, we could reduce the random thermal motions of the atoms, making them easy to trap. Second, we could easily load the atoms into the trap. Simply by focusing the trapping beam in the center of the optical molasses, atoms would be snagged as they randomly wandered into the trapping beam. Third, by alternating between trapping and cooling light, we could reduce the heating effects of the trapping light. A year after we had perfected optical molasses, atoms could finally be trapped with light.
Even with the loading technique used in our first trap, an optical trap with a larger capture volume was desirable. A trap that could use the scattering force would need much less light intensity, which meant the constraints imposed by the Optical Earnshaw Theorem had to be circumvented. The important clue of how to design such a trap came from David E. Pritchard of the Massachusetts Institute of Technology and Carl E. Wieman of the University of Colorado and their colleagues. They pointed out that if magnetic or electric fields that varied over space were applied to atoms, the scattering force caused by the laser light would not necessarily be proportional to the light intensity.
This suggestion led Jean Dalibard of the Ecole Normale Superieure in Paris to propose a "magneto-optic" trap, which used a weak magnetic field and circularly polarized light. In 1987 Pritchard's group and my own at AT&T collaborated to construct such a trap. Three years later Wieman's team went on to show that this technique could be used to trap atoms in a glass cell, using inexpensive diode lasers. Their method eliminated the precooling procedures needed in our first trapping experiments. The fact that atoms could be trapped in a sealed cell also meant rare species of atoms, such as radioactive isotopes, could be optically manipulated. The magneto-optic trap has become the most widely used optical trap today.
Meanwhile researchers were making rapid progress in laser cooling. Phillips and his colleagues discovered that under certain conditions, optical molasses could be used to cool atoms to temperatures far below the lower limit predicted by the existing theory. This discovery prompted Dalibard and Claude Cohen-Tannoudji of the College de France and the Ecole Normale and my group at Stanford to construct a new theory of laser cooling based on a complex but beautiful interplay between the atoms and their interaction with the light fields. Currently atoms can be cooled to a temperature with an average velocity equal to three and a half photon recoils. For cesium atoms, it means a temperature lower than three microkelvins.