Construct a dark state and test its limits

Consider a Lambda atom with two ground states g and s and one excited state e. A probe drives g to e and a control drives s to e. In a rotating frame at two-photon resonance, choose their Rabi frequencies real and positive for convenience. The interaction contains both excitation amplitudes with a common one-half convention. A superposition of g and s is dark when those two amplitudes add to zero. This is a statement about amplitudes before probabilities are squared.

The normalized dark state has ground-state coefficients proportional to control strength and minus probe strength. It has no excited-state component, so ideal spontaneous decay from e cannot act on it. Equal couplings produce equal ground populations with opposite phase. Changing their ratio rotates the dark state. This exact eigenvector statement does not itself prove that an arbitrary initial atom is already in that state, that it is prepared instantaneously, or that a finite bandwidth pulse propagates without loss.

For susceptibility, assume the control is present, the probe is weak enough that nearly all population remains in g, and the response is stationary. The optical coherence couples to the ground coherence. Eliminating the latter gives a probe response whose numerator contains ground-state decoherence minus i times two-photon detuning. On exact Raman resonance with vanishing ground decoherence, the numerator vanishes while a nonzero control keeps the denominator finite. The absorptive response therefore vanishes ideally.

This weak-probe derivation and the exact dark-state formula serve different purposes. The dark eigenstate exists for comparable coherent couplings, whereas the linear susceptibility assumes the probe is small compared with the control and does not appreciably redistribute population. A control can also split an upper-state resonance, and experimentally separating interference-based narrow transparency from a resolved dressed-state splitting requires comparing rates and spectra. The label transparency alone does not determine which parameter regime produced a measured feature.

At exact one- and two-photon resonance, the remaining optical response is purely absorptive in this model and scales with ground dephasing. Relative to the same weak probe without the control, its absorption ratio is the product of the two decoherence rates divided by that product plus control strength squared over four. Stronger control suppresses this residual response, but also changes the transparency bandwidth and dispersion. Motion, collisions, magnetic noise, and laser phase fluctuations can degrade the necessary ground coherence.

A transparent narrow spectral window can have steep dispersion and a small group velocity, yet a finite pulse must fit that window and retain suitable adiabaticity and optical depth. That does not imply instantaneous information storage or superluminal information transmission. The pure two-state lab represents only the ground-subspace geometry of a dark state. It has no third level, probe susceptibility, dissipative preparation, or propagation. Specify both angles and compare superposition phases without calling its precession a transparency simulation.