Kilohertz-Resolution Spectroscopy of Cold Atoms with an Optical Frequency Comb

We have performed sub-Doppler spectroscopy on the narrow intercombination line of cold calcium atoms using the amplified output of a femtosecond laser frequency comb. Injection locking of a 657-nm diode laser with a femtosecond comb allows for two regimes of amplification, one in which many lines of the comb are amplified, and one where a single line is predominantly amplified. The output of the laser in both regimes was used to perform kilohertz-level spectroscopy. This experiment demonstrates the potential for high-resolution absolute-frequency spectroscopy over the entire spectrum of the frequency comb output using a single high-finesse optical reference cavity.

Figure 1

Experimental depiction for high-precision optical spectroscopy of cold Ca atoms using a femtosecond laser frequency comb. Acousto-optic modulators (AOM) switch light between two optical fibers that deliver counterpropagating 657-nm pulses to the Ca magneto-optical trap (MOT) (the 423-nm shelving detection beams are left out for simplicity). An $f\mathrm{\text{-to-}}2f$ self-referencing technique allows phase stabilization of the laser offset frequency, $f_0$. Phase stabilization of the laser repetition rate, $f_{\mathrm{rep}}$, is obtained by phase locking the comb to a fiber laser at 1068 nm that is stabilized to a high-finesse optical cavity. Measurement of the laser repetition rate vs a hydrogen maser permits for frequency calibration in the measurements.

Figure 2

Relevant ${}^{40}\mathrm{Ca}$ energy levels.

Figure 3

Saturated absorption dip observed on the Doppler profile of the Ca clock transition using two counterpropagating pulses from a comb-injected slave laser with a current of (a) 65 mA and (b) 98 mA. The double peak observed in the case of low power broadening [case (a)] is the recoil doublet. In case (b), amplification of preferred comb lines leads to greater power broadening ($\mathrm{\text{linewidth}}=108\mathrm{kHz}$), thereby making the double structure indistinguishable. Each point in the plots is the result of 150 ms of averaging. The $y$ axis is offset for convenience and 1 mV corresponds to $\sim 600$ atoms, which is estimated from the measured fractional excitation and the trapped atom number.

Figure 4

Time-resolved optical Bordé-Ramsey fringes for pulse lengths of $7\mu \mathrm{s}$ and varying pulse separations of (a) $14\mu \mathrm{s}$, (b) $104\mu \mathrm{s}$, and (c) $203\mu \mathrm{s}$. The decrease in the signal-to-noise ratio from (a) to (c) is the result of spontaneous population decay from the excited state and residual noise on the 657-nm light (see text for details). Each point in the plots is the result of 150 ms of averaging. The $y$ axis is offset for convenience and 1 mV corresponds to $\sim 600$ atoms, which is estimated from the measured fractional excitation and the trapped atom number.