Rydberg atom radio-frequency field sensor (2D)

Side view of the vapour cell with the EIT spectrum, dressed-state ladder and the field read-out computed from the splitting.

Switch to 3D

Presets

Playback

RF field and Rydberg level

Lasers and vapour

Probe scan

Layers

Model check

The science

What a Rydberg-atom sensor measures

Rubidium vapour sits in a glass cell. A weak probe laser (780 nm) drives 5S to 5P and a strong coupling laser (480 nm) drives 5P to a Rydberg level nD, travelling the opposite way. Together they make the vapour transparent in a narrow window: electromagnetically induced transparency (EIT). A radio-frequency field tuned to a neighbouring Rydberg transition mixes the two Rydberg states, and the single transparency peak splits in two (Autler-Townes splitting). The gap between the peaks is proportional to the field amplitude, and the constant of proportionality is an atomic dipole moment and Planck's constant, so the reading needs no calibration against a reference antenna.

Equations used

Weak-probe steady state of the ladder g - e - r - r' (all rates and detunings in units of 2 pi MHz):

rho_ge = (i Op/2) / D ,   absorption ~ Re(1/D)
D = g_e - i d1 + (Wc/2)^2 / [ g_r - i d2 + (Wrf/2)^2 / (g_r' - i d2) ]
d1 = dp - u ,   d2 = dp + u (lambda_p/lambda_c - 1)       (u = k_p v / 2 pi)
Wrf / 2 pi = mu E / h ,   mu = c n^2 e a0 ,   c = 0.35
splitting on the probe axis = (lambda_c / lambda_p) Wrf / 2 pi
T = exp(-OD) ,   OD(dp) = sum over velocity classes with Maxwell weights
  • Doppler averaging is done numerically: every velocity class of the 300 K vapour (sigma = 217 MHz on the probe line) is summed. The factor lambda_c / lambda_p = 0.615 is not put in by hand; it appears because the two Doppler shifts do not cancel for the two-photon detuning.
  • mu scales as n squared. The Rabi frequency per volt per metre therefore grows as n squared and the smallest resolvable field falls as 1/n squared. The constant c is a representative value for neighbouring Rydberg states, not a specific transition; real sensors use a calculated matrix element.
  • Resolvability. The two peaks are separate only if the splitting exceeds roughly twice the EIT window width. Below that the read-back is biased high and the page says the peaks have merged.

Honest simplifications

Three-level weak-probe theory with a single RF-coupled neighbour; no magnetic sublevels, no hyperfine structure, no laser linewidth beyond one lumped Rydberg decoherence rate, no propagation along the cell (one optical depth, OD = 3 at the bare line centre). Laboratory sensors reach micro-volt-per-centimetre sensitivity with heterodyne read-out, which is not modelled here; this page shows the direct splitting regime.

Independent numerical checks (shown in the Model check list)

  1. The closed-form continued fraction against a Gaussian-elimination solution of the three coupled steady-state equations.
  2. The splitting found in the computed spectrum against (lambda_c / lambda_p) mu E / h at a well-resolved field.
  3. Convergence of the velocity integral when the step is halved.
  4. The n squared scaling of the dipole and the identity Rabi frequency = mu E / h.
Probe transmission spectrum (Doppler-averaged) with the predicted peaks
Peak splitting against RF field: prediction and spectrum
Smallest resolvable field against Rydberg level n
Complex susceptibility (Argand plot) along the scan
Made with 🟠 Blender · Three.js