We describe new methods to consistently isolate and image individual atoms. We isolate one atom by inducing inelastic light assisted collisions in a group of atoms held in optical tweezers. Each atom pair that undergoes a collision gains enough energy for only one of the atoms to leave the optical tweezers, leading to a single atom remaining. This atom is imaged using fluorescence microscopy.

Optical fluorescence microscopy of individual neutral atoms has become an increasingly popular tool in atomic physics during the past decade [1-4]. It is employed both as a tool in attempts to make neutral atom quantum logic devices [2, 5] and to study many-body physics at the single atom level [3, 4].

Experiments involving fluorescence microscopy of individual neutral atoms present a number of challenges. First, the atoms have to be held in place long enough to collect the required fluorescence for an image to be formed. At the same time, we need to eliminate stray light from the imaging system. Finally, the atoms have to be isolated from each other such that they can be resolved by the microscope. Here we describe a novel scheme to consistently isolate and image individual atoms on a sub-second time-scale.

Isolating Individual Atoms

Preparation of a single neutral atom in the focal plane of the microscope is a three-step process [6]. The first step is to slow down a group of atoms inside an Ultra-High Vacuum (UHV) chamber. Due to their low mass, atoms at room temperature move rapidly, with typical speeds of several hundred meters per second. To slow down atoms from a low flux beam, we employ laser cooling techniques. These techniques are based on the fact that an atom recoils when it absorbs or emits a photon. Through this the light can exert a force on the atom, and we make use of a configuration of laser beams and a magnetic field such that this force opposes the motion of the atom regardless of the direction in which it is travelling [7]. This configuration is known as a Magneto Optical Trap (MOT) and it produces a cloud of about 100,000 85Rb atoms at a temperature around 1 mK, which correspond to a typical atomic speed of ~10 cm/s.

The process is illustrated in figure 2a.

In the second step we load a fraction of these laser-cooled atoms into ­„optical tweezers". The optical tweezers is a focused laser beam with a frequency significantly less than the lowest resonance in the atoms. This beam can hold the atoms in its focus due to the difference in index of refraction between the atoms and the surrounding vacuum [8]. We form optical tweezers of transverse size ω0 = 1.8 μm inside the MOT cloud by passing a laser beam of wavelength 828 nm through a 0.55 NA objective lens. The volume of the focus of the optical tweezers beam is significantly smaller than the atomic MOT cloud, so that only about 50 atoms end up held by the optical tweezers. At this point, the laser cooling is turned off and the rest of the atoms fall to the bottom of the vacuum chamber, as illustrated in figure 2b.

The final step in producing a single atom uses "light assisted collisions" to expel all atoms but one from the optical tweezers. To this end, we direct a laser beam with tailored frequency and intensity onto the group of atoms held by the optical tweezers (see Fig. 2c). The light makes the atoms repel each other and gain energy, such that the atoms will be moving faster after a collision than before. In this inelastic process, the gained energy is supplied by the light. By careful choice of the parameters of the collision inducing light, we can limit the energy released in a collision to the amount required for one of the atoms to escape the optical tweezers [6]. Each inelastic collision therefore results in one of the collision partners leaving the optical tweezers while the other stays. This process, where the atoms "kick each other out", continues until only one atom is left in the optical tweezers.

To check the efficiency of the scheme, we run this procedure 1,000 times and at the end of each run take a fluorescence image of the atoms in the optical tweezers as described below. Figure 3 is a histogram of the amount of light scattered by the atoms in the optical tweezers and shows discrete peaks corresponding to zero or one atom in the optical tweezers and a very small peak for two atoms. 82.7% of the runs ended with one atom in the optical tweezers. There are several reasons why the scheme in practice does not work 100% of the time. For instance, in some runs where a single atom is prepared, it is lost before detection. A second contribution to this comes from a small probability of losing a pair due to light assisted collisions [6].