Optical nonlinearities offer unique possibilities for the control of light with light. A prominent example is electromagnetically induced transparency (EIT), where the transmission of a probe beam through an optically dense medium is manipulated by means of a control beam1, 2, 3. Scaling such experiments into the quantum domain with one (or just a few) particles of light and matter will allow for the implementation of quantum computing protocols with atoms and photons4, 5, 6, 7, or the realization of strongly interacting photon gases exhibiting quantum phase transitions of light8, 9. Reaching these aims is challenging and requires an enhanced matter–light interaction, as provided by cavity quantum electrodynamics10, 11, 12. Here we demonstrate EIT with a single atom quasi-permanently trapped inside a high-finesse optical cavity. The atom acts as a quantum-optical transistor with the ability to coherently control13 the transmission of light through the cavity. We investigate the scaling of EIT when the atom number is increased one-by-one. The measured spectra are in excellent agreement with a theoretical model. Merging EIT with cavity quantum electrodynamics and single quanta of matter is likely to become the cornerstone for novel applications, such as dynamic control of the photon statistics of propagating light fields14 or the engineering of Fock state superpositions of flying light pulses15.
Figures at a glance
Figure 1: Experimental protocol and cavity EIT. 
87Rb atoms are quasi-permanently trapped inside a high-finesse optical cavity. The cavity is resonant with the atomic F = 1 ↔ F′ = 1 transition at 780 nm wavelength. The transmission of the atom–cavity system is probed with a weak laser (probe laser–cavity detuning Δ) for three physical conditions. a, With atoms shelved in the hyperfine state F = 2, we record the empty cavity transmission as a reference (black data and curve in d). b, With atoms prepared in F = 1, we realize a cavity QED situation and observe a spectrum exhibiting a vacuum-Rabi splitting (red data and curve in d). c, An additional laser is used to coherently control the optical properties of the atom–cavity system. d, Measured transmission spectra for on average 15 atoms coupled to the cavity. A narrow transmission window (linewidth ~900 kHz) observed at the two-photon resonance in the cavity EIT situation (blue data and theory curve) testifies to the existence of a coherent dark state. Experimental parameters: maximum intra-cavity photon number 0.02, control power 9 µW (equivalent Rabi frequency 1.3κ). Error bars shown are ±s.d. and are omitted from the empty cavity and two-level measurement for clarity.
Figure 2: Cavity EIT with a single atom. 
Colour coding same as in Fig. 1d. a, Measured transmission spectra for exactly one atom coupled to the cavity and a control laser power of 3 µW (equivalent Rabi frequency, 0.78κ). EIT is observed with a maximum transparency of 96% and a measured transmission contrast of 20% with respect to the control laser switched off. The linewidth (see Methods Summary) is ~1.2 MHz. The red solid curve is a solution of the time-dependent master equation for the finite probing interval. The red dashed curve is the prediction of equation (1) for zero control power. Error bars shown are ±s.d. Inset, CCD camera image of a single atom trapped in the cavity (image size, 33 µm × 16 µm). b, c, The linewidth and contrast of the single-atom transparency feature are tunable by means of the control laser power. The values used are (1, 2, 3) µW and correspond to Rabi frequencies (0.45, 0.63, 0.78) κ. The maximum intra-cavity photon number is 0.02. Error bars shown are ±s.d. for the contrast, ±0.03κ for the linewidth and ±10% for the x axis.
Figure 3: Cavity EIT spectra for N = 2 to 5 atoms. 
a, 2 atoms; b, 3 atoms; c, 4 atoms; d, 5 atoms. Changing the number of atoms enhances the visibility of the dark-state resonance owing to improved contrast with respect to the measurements with two-level atoms coupled to the cavity. The vacuum-Rabi splitting starts being resolved at higher atom number. Colour coding and experimental parameters are as in Fig. 2a. Insets, CCD camera images of the trapped atoms, used to precisely determine their number and physical location in the cavity mode. Error bars shown are ±s.d. and are omitted from the empty cavity and two-level measurement for clarity.
Figure 4: Measured transparency, contrast and linewidth of cavity EIT with N = 1 to 7 atoms. 
a, The maximum transparency (blue bars, top) decreases with the number of atoms from 96% (N = 1) to 78% (N = 7). Nevertheless, the on/off contrast (red bars, bottom) at the two-photon resonance steadily increases from 21% (N = 1) to 60% (N = 7) owing to the reduction of transmission with control laser switched off. b, For N ≥ 3, the cavity EIT linewidth decreases with the number of coupled atoms as 1/N (guide to the eye, dashed red curve). For N = 1, 2, the linewidth is nearly constant owing to the interplay between the increase in the two-level atom transmission and the definition of the linewidth used (see Methods Summary). Error bars shown are ±s.d. for the contrast and transparency and ±0.03κ for the linewidth.
right
