In laser cooling, it's my understanding that the lasers are slightly detuned from the absorption frequency of the atoms so that only atoms moving a specific direction absorb light from a given laser. Once an atom absorbs a photon, it excites an electron to transition up in energy level. Then, the electron transitions back down and emits a photon. This photon has momentum and imparts an equal and opposite momentum to the atom to conserve momentum.
If this is (roughly) correct, I would expect the emission to be spontaneous and therefore random in direction. If the direction a photon is emitted after it is absorbed by an atom is random, how can this approach work to slow atoms? I would expect the atom to lose momentum $50\%$ of the time and gain momentum the other $50\%$.
I've read here that it is inaccurate to view the emission of the photon as having a particular "direction" but this would seem to invalidate my explanation for how atoms are cooled (viewing them as particles with definite directions of motion and therefore vector quantities of momentum).
Where is my misconception?
