Normal-Mode Spectroscopy of a Single-Bound-Atom–Cavity System

The energy-level structure of a single atom strongly coupled to the mode of a high-finesse optical cavity is investigated. The atom is stored in an intracavity dipole trap and cavity cooling is used to compensate for inevitable heating. Two well-resolved normal modes are observed both in the cavity transmission and the trap lifetime. The experiment is in good agreement with a Monte Carlo simulation, demonstrating our ability to localize the atom to within $\lambda /10$ at a cavity antinode.

Figure 1

Experimental setup: The high-finesse cavity is excited by a weak near-resonant probe field and a strong far-red-detuned trap field. ${}^{85}\mathrm{R}\mathrm{b}$ atoms are injected from below. Behind the cavity, the two light fields are separated by a grating and measured with independent photodetectors.

Figure 2

Transmission of the cavity containing a single trapped and strongly coupled atom (circles). The detuning between the cavity and the atom is adjusted by tuning the Stark shift of the atom via the trapping-field power expressed in terms of the transmitted power, $P$. The average transmission during probe intervals for which the atom is found to be strongly coupled by independent qualification (see text) shows well-resolved normal-mode peaks. On average each point includes the data from about 350 probe intervals collected from between $35$ and $1000$ atoms. A Monte Carlo simulation (solid lines) describes the data well.

Figure 3

Simulated probability distribution of the atom-cavity coupling for all probe intervals (dotted line) and for the strongly coupled intervals (solid line). Qualification completely eliminates the occurrence of probe intervals with weak coupling, $g⪅(\gamma ,\kappa )$.

Figure 4

Probe-induced loss rate of atoms from the trap (triangles) for different detunings between cavity and atom, adjusted by varying the trapping-field power. No qualification is employed. The experiment is in qualitative agreement with a Monte Carlo simulation (solid lines): the observed frequencies, widths, and relative heights of the normal-mode peaks are well described by the simulation. Only the absolute value of the measured rate exceeds that of the simulation. This could be explained by fluctuations of experimental parameters not taken into account in the simulation, e.g., the lack of shot noise in the modeling of atom capture. For all probe detunings the simulated atomic excitation is below $1.4\%$.