Rabi Frequency and Generalized Rabi Frequency

Take the transition dipole moment and the electric-dipole approximation from the two articles before this one, then ask a sharper question: not how much light a two-level atom absorbs on average, but what the amplitude does, instant by instant, while a coherent field is switched on. The answer is not the monotone climb of the Einstein rate picture — the amplitude sloshes, the atom cycling from ground to excited and back for as long as phase coherence survives. The angular frequency of that oscillation is the Rabi frequency, the single most important number in coherent light–matter physics, and it is nothing more than the dipole interaction energy in frequency units: $\hbar\Omega$ is (up to sign) the coupling $-\mathbf{d}\cdot\mathbf{E}_0$ . A bigger dipole or stronger field means faster flopping. That reading rests on two approximations from earlier in the chapter — the field is uniform across the atom (electric dipole approximation), and near resonance the fast counter-rotating term at $\omega+\omega_0$ is dropped (rotating-wave approximation). Two things then organize everything. First, detuning by $\delta=\omega-\omega_0$ speeds the flopping up to the generalized Rabi frequency $\tilde\Omega=\sqrt{\Omega^2+\delta^2}$ while shrinking its amplitude to $\Omega^2/\tilde\Omega^2$ — an atom driven off resonance can never be fully inverted. Second, spontaneous emission at rate $\Gamma$ damps the oscillation: when $\Omega\gg\Gamma$ the atom rings for many cycles; when $\Gamma$ dominates, the flopping washes out into the incoherent rate equations. The Rabi frequency is the hinge between the coherent and incoherent descriptions of an atom in light.

Key takeaways

The Rabi frequency $\Omega=-\mathbf{d}_{eg}\cdot\mathbf{E}_0/\hbar$ is the angular rate at which a coherent field drives population between two levels — set by the transition dipole and field amplitude, not the optical frequency.
On resonance the excited-state probability is $\sin^2(\Omega t/2)$ : a full-contrast flop between $0$ and $1$ , fundamentally unlike incoherent rate-equation absorption.
Off resonance the dynamics run at the generalized Rabi frequency $\tilde\Omega=\sqrt{\Omega^2+\delta^2}$ : faster, but capped at a maximum population $\Omega^2/\tilde\Omega^2<1$ .
The pulse area $\theta=\int\Omega\,\mathrm{d}t$ is a rotation angle — $\pi$ inverts, $\pi/2$ superposes — the basis of qubit gates, Ramsey spectroscopy, and coherent control.
Spontaneous emission at rate $\Gamma$ damps the flopping into a steady state that saturates at $1/2$ ; the saturation parameter $s=I/I_{\text{sat}}$ ties $\Omega$ to intensity.
Key idea: in the coherent two-level limit, a near-resonant field makes the population oscillate at $\tilde\Omega=\sqrt{\Omega^2+\delta^2}$ ; the on-resonance amplitude is set by the dipole coupling $\Omega=-\mathbf{d}\cdot\mathbf{E}_0/\hbar$ , and the resonant pulse area $\int\Omega\,\mathrm{d}t$ fixes the final state.

The core insight

A coherent near-resonant field makes the two-level population oscillate at $\tilde\Omega=\sqrt{\Omega^2+\delta^2}$ ; the on-resonance amplitude is set by the dipole coupling $\Omega=-\mathbf{d}\cdot\mathbf{E}_0/\hbar$ , and the pulse area $\int\Omega\,\mathrm{d}t$ controls where on the Bloch sphere the state ends up.

Where this lives today

The Rabi frequency is the clock rate of quantum technology. In a trapped-ion or superconducting processor a single-qubit gate is a calibrated pulse area: engineers dial $\Omega$ through the drive amplitude and set its phase to pick the rotation axis, so a short pulse executes a NOT gate ( $\pi$ ) or a $\pi/2$ rotation. Fidelity hinges on knowing $\Omega$ to a percent and keeping $\Omega\gg\Gamma$ . The same physics runs the world's clocks: an optical-lattice or fountain standard interrogates its atoms with a Ramsey sequence — two $\pi/2$ -pulses bracketing a long free evolution — whose fringes narrow with interrogation time. It also shapes everyday AMO practice. Laser cooling relies on the saturated scattering rate $\Gamma\rho_{ee}^{ss}$ , set by $s=2\Omega^2/\Gamma^2$ , with power broadening kept in check by modest $s$ . Dipole traps and optical tweezers use a red-detuned AC Stark shift $\Delta_{\rm Stark}\approx\Omega^2/(4\delta)<0$ to create a potential well holding single atoms for neutral-atom arrays. And under strong driving the Rabi frequency appears in the frequency domain as the Mollow triplet and the Autler–Townes doublet split by $\tilde\Omega$ .

The physical picture

Model the two-level atom as a spin-½ on the Bloch sphere: south pole the ground state, north pole the excited state, points between are superpositions. A near-resonant field acts on this pseudospin exactly like a magnetic field on a real spin — a torque, about which the state precesses at a rate equal to the effective vector's length $\tilde\Omega$ . That single fact contains the whole subject. On resonance the torque lies flat in the equator with length $\Omega$ , and the state precesses on a great circle over both poles: from the south pole it sweeps all the way to the north (full inversion) and back. Detuning tilts the vector up by $\arctan(\delta/\Omega)$ and lengthens it to $\tilde\Omega$ , so the state precesses on a cone that never reaches the north pole (population capped below one) and, being longer, precesses faster. Reduced contrast and quicker flopping are read straight off the geometry.

Figure

Effective-field geometry of the driven two-level atom. On resonance the torque lies in the equator and the state sweeps ground to excited on a great circle; a detuning δ tilts the effective vector and lengthens it to the generalized Rabi frequency

\tilde\Omega

, so the state precesses on a cone that misses the excited pole. For δ = Ω the tilt is 45° and the cone tops out at half inversion.

The flopping is a coherent phenomenon, separating it from the Einstein rate-equation picture of absorption under broadband light, where the population climbs monotonically because the random phase of the radiation continually scrambles the dipole. Rabi flopping is what you see before spontaneous emission and collisions erase the phase. The second figure plots it for a realistic optical Rabi frequency, contrasting the resonant full flop with the reduced, faster oscillation at a detuning equal to $\Omega$ .

Figure

Coherent Rabi flopping of the excited-state population for a drive with Ω over 2π = 10 MHz, so the on-resonance period is 2π over Ω = 100 ns. The resonant curve (blue) reaches full inversion of

1

; a detuning

\delta = \Omega

(amber) oscillates faster, at

\tilde{\Omega} = \sqrt{2}\,\Omega

, but tops out at only

\Omega^2

over

\tilde{\Omega}^2 = 0.5

The pulse area $\theta=\int\Omega\,\mathrm{d}t$ is the angle the state rotates through: $\pi$ flips ground to excited, $2\pi$ returns it (with a sign flip), $\pi/2$ makes an equal superposition. This is the language of NMR spin echoes and quantum-logic gates — controlling $\Omega$ and its phase is a complete description of single-qubit control.

The core equations

Everything starts from the coupling energy: the Rabi frequency is the transition dipole dotted into the field amplitude, divided by $\hbar$ so it is in angular-frequency units — the interaction energy as a frequency.

\Omega = -\frac{\mathbf{d}_{eg}\cdot\mathbf{E}_0}{\hbar} \qquad (1)

where: $\Omega$ — Rabi frequency (angular, $\mathrm{rad\,s^{-1}}$ ); $\mathbf{d}_{eg}=\langle e|(-e\mathbf{r})|g\rangle$ — transition dipole matrix element; $\mathbf{E}_0$ — field amplitude, $\mathbf{E}(t)=\mathbf{E}_0\cos\omega t$ ; $\hbar$ — reduced Planck constant. Reading: $\Omega\propto E_0\propto\sqrt I$ . Off resonance the atom responds to the combination of coupling and detuning. The generalized Rabi frequency is the length of the effective Bloch vector — the quadrature sum of $\Omega$ and $\delta$ .

\tilde\Omega = \sqrt{\Omega^2 + \delta^2},\qquad \delta = \omega - \omega_0 \qquad (2)

where: $\tilde\Omega$ — generalized Rabi frequency; $\delta$ — detuning; $\omega$ — drive frequency; $\omega_0=(E_e-E_g)/\hbar$ — transition frequency. Reading: on resonance $\tilde\Omega=\Omega$ ; any detuning only increases the flopping rate. Solving in the rotating frame gives the full time dependence — a sine-squared oscillation whose rate is $\tilde\Omega$ and whose amplitude is throttled by $\Omega^2/\tilde\Omega^2$ . This is the Rabi formula.

P_e(t) = \frac{\Omega^2}{\tilde\Omega^2}\,\sin^2\!\left(\frac{\tilde\Omega\,t}{2}\right) \qquad (3)

where: $P_e(t)$ — excited-state probability at time $t$ ; $\Omega^2/\tilde\Omega^2$ — peak population, a Lorentzian in $\delta$ of half-width $\Omega$ . Reading: on resonance the prefactor is $1$ and the atom fully inverts; off resonance it drops. The outcome of a pulse depends only on its area — the time integral of $\Omega$ , which on resonance is the Bloch-sphere rotation angle. Two areas are the elementary gates of coherent control.

\theta=\int_0^t\Omega(t')\,\mathrm{d}t';\qquad \theta=\pi\Rightarrow P_e=1,\quad \theta=\tfrac{\pi}{2}\Rightarrow P_e=\tfrac12 \qquad (4)

where: $\theta$ — pulse area; $\pi$ inverts, $\pi/2$ makes an equal superposition. Reading: for constant $\Omega$ , $t_\pi=\pi/\Omega$ . Far off resonance the atom is barely excited but not untouched: the field pushes its levels around. This AC Stark (light) shift is the second-order dressed-level energy, scaling as intensity over detuning.

\Delta_{\text{Stark}} \approx \frac{\Omega^2}{4\delta}\qquad(|\delta|\gg\Omega) \qquad (5)

where: $\Delta_{\rm Stark}$ — ground-state AC Stark shift; $\delta$ — (large) detuning. Reading: for red detuning, $\Delta_{\rm Stark}<0$ , so intensity maxima lower the ground-state energy and act as potential wells — a trap. With spontaneous emission the oscillation settles into a steady state, written with the saturation parameter $s$ ; the steady population never exceeds one half.

\rho_{ee}^{ss}=\frac{1}{2}\,\frac{s}{1+s},\qquad s=\frac{\Omega^2/2}{\delta^2+(\Gamma/2)^2} \qquad (6)

where: $\rho_{ee}^{ss}$ — steady-state excited population; $s$ — saturation parameter; $\Gamma=1/\tau$ — decay rate. Reading: on resonance $s=2\Omega^2/\Gamma^2=I/I_{\text{sat}}$ , and $\rho_{ee}^{ss}\to1/2$ as $s\to\infty$ . In the frequency domain the same drive splits each level into dressed states separated by $\hbar\tilde\Omega$ — the Autler–Townes doublet, the spectroscopic fingerprint of Rabi flopping.

Derivation AFrom the dipole interaction to the rotating-wave Hamiltonian

The goal is to see the Rabi frequency emerge as the off-diagonal coupling of the two-level Hamiltonian, then turn the time-dependent problem into a static one by moving to a rotating frame and dropping the fast counter-rotating term. The payoff is a two-by-two matrix whose eigenvalue splitting is exactly $\hbar\tilde\Omega$ , so the generalized Rabi frequency is the energy gap of the driven system, not an ad-hoc definition. We start from a two-level atom in a classical field and keep only the resonant term.

Step 1. Write the Hamiltonian as bare atom plus dipole coupling: $H=\hbar\omega_0\,|e\rangle\langle e|-\mathbf{d}\cdot\mathbf{E}_0\cos(\omega t)$ , with the ground energy zero. Why: the electric dipole approximation replaces the interaction with $-\mathbf{d}\cdot\mathbf{E}$ , valid because the optical wavelength dwarfs the atom, so the field is uniform across it.

Step 2. The coupling lives off-diagonal: $\langle e|H|g\rangle=-\mathbf{d}_{eg}\cdot\mathbf{E}_0\cos\omega t\equiv\hbar\Omega\cos\omega t$ , defining $\Omega=-\mathbf{d}_{eg}\cdot\mathbf{E}_0/\hbar$ . Why: only $\mathbf{d}_{eg}$ connects the two levels, so the interaction that drives population is captured by this one number — the Rabi frequency.

Step 3. Split the cosine, $\cos\omega t=\tfrac12(e^{i\omega t}+e^{-i\omega t})$ , and move to the interaction picture; the coupling then carries a slow beat at $\omega-\omega_0=\delta$ and a fast one at $\omega+\omega_0$ . Why: separating the exponentials exposes which part of the drive stays in step with the atom (co-rotating) and which fights it (counter-rotating).

Step 4. Make the rotating-wave approximation: drop the $e^{\pm i(\omega+\omega_0)t}$ term. Why: near resonance it oscillates at roughly $2\omega_0$ , far faster than the dynamics, so it averages to nothing whenever $\Omega,|\delta|\ll\omega_0$ .

Step 5. Transformed into the frame rotating at $\omega$ , what remains is the static $H_{\text{RWA}}=-\tfrac{\hbar\delta}{2}\sigma_z+\tfrac{\hbar\Omega}{2}\sigma_x$ , with eigenvalues $\pm\tfrac{\hbar}{2}\sqrt{\Omega^2+\delta^2}=\pm\tfrac{\hbar\tilde\Omega}{2}$ . Why: diagonalizing a matrix with diagonal $\mp\delta/2$ and off-diagonal $\Omega/2$ gives a splitting equal to the quadrature sum — $\tilde\Omega$ is literally the dressed-level gap.

Go deeper · Derivation B · 1 minSolving the rotating-frame amplitudes for the Rabi formula

Derivation BSolving the rotating-frame amplitudes for the Rabi formula

The goal is to turn the static rotating-wave Hamiltonian into the time evolution of the excited-state probability, recovering the Rabi formula $P_e=(\Omega^2/\tilde\Omega^2)\sin^2(\tilde\Omega t/2)$ . Both headline features — a rate set by $\tilde\Omega$ and an amplitude capped by $\Omega^2/\tilde\Omega^2$ — fall out of one second-order equation, starting from the atom in the ground state.

Step 1. Schrödinger's equation in the rotating frame gives the coupled pair $i\dot c_e=-\tfrac{\delta}{2}c_e+\tfrac{\Omega}{2}c_g$ and $i\dot c_g=\tfrac{\delta}{2}c_g+\tfrac{\Omega}{2}c_e$ . Why: these are $i\dot{\mathbf c}=H_{\text{RWA}}\mathbf c/\hbar$ component by component, with the fast oscillation already removed.

Step 2. Differentiate and substitute to eliminate $c_g$ , giving one oscillator equation $\ddot c_e+\tfrac14(\Omega^2+\delta^2)c_e=0$ . Why: decoupling into a second-order equation makes the natural frequency visible — $\tilde\Omega/2$ , so the amplitude beats at $\tilde\Omega/2$ and the probability at $\tilde\Omega$ .

Step 3. Impose $c_e(0)=0$ (atom in the ground state), which selects $c_e(t)=-i\,\tfrac{\Omega}{\tilde\Omega}\sin\!\big(\tfrac{\tilde\Omega t}{2}\big)$ . Why: the excited amplitude is fixed by $\Omega/\tilde\Omega$ , so any detuning making $\tilde\Omega>\Omega$ shrinks the amplitude the excited state can reach.

Step 4. Take the modulus squared: $P_e=|c_e|^2=\tfrac{\Omega^2}{\tilde\Omega^2}\sin^2\!\big(\tfrac{\tilde\Omega t}{2}\big)$ . Why: squaring gives the sine-squared population and fixes the peak at $\Omega^2/\tilde\Omega^2$ — unity on resonance, less off it, so a detuned atom cannot fully invert.

Go deeper · Derivation C · 1 minAdding decay: the optical Bloch equations and saturation

Derivation CAdding decay: the optical Bloch equations and saturation

The goal is to graft spontaneous emission onto the coherent picture and derive the saturated steady-state population, showing why an atom driven arbitrarily hard still spends at most half its time excited. Amplitudes cannot describe an open system, so we promote them to a density matrix with decay rate $\Gamma$ for the population and $\Gamma/2$ for the coherence, then set the derivatives to zero for the steady state that spectroscopy and cooling measure.

Step 1. The population equation acquires a loss term: $\dot\rho_{ee}=-\Gamma\rho_{ee}+\tfrac{i\Omega}{2}(\rho_{eg}-\rho_{ge})$ . Why: $-\Gamma\rho_{ee}$ is spontaneous emission draining the excited state, the irreversible ingredient absent from the Schrödinger treatment.

Step 2. The coherence decays and oscillates: $\dot\rho_{eg}=-(\tfrac{\Gamma}{2}-i\delta)\rho_{eg}+\tfrac{i\Omega}{2}(\rho_{ee}-\rho_{gg})$ . Why: it dephases at half the population rate, and that dephasing destroys clean flopping and forces a steady state.

Step 3. Set every derivative to zero and solve for $\rho_{ee}$ : $\rho_{ee}^{ss}=\dfrac{\Omega^2/4}{\delta^2+\Gamma^2/4+\Omega^2/2}$ . Why: under continuous driving the atom balances coherent pumping against decay — the population a probe or cooling laser sees.

Step 4. Read the limits: on resonance at strong drive $\rho_{ee}^{ss}\to\tfrac12$ ; at weak drive it is a Lorentzian of full width $\Gamma$ in $\delta$ , broadening as $\Gamma\sqrt{1+s}$ . Why: the half ceiling is stimulated emission catching up with absorption, and the broadening is power broadening — why saturated linewidths overstate the natural width.

Limiting cases

Case	Result	Meaning
Exact resonance	$\delta=0\Rightarrow\tilde\Omega=\Omega,\ P_e=\sin^2(\Omega t/2)$	Full-contrast flop $0$ to $1$ ; inversion at $t=\pi/\Omega$ .
Large detuning	$\lvert\delta\rvert\gg\Omega\Rightarrow\tilde\Omega\approx\lvert\delta\rvert,\ P_e^{\max}=\Omega^2/\delta^2\ll1$	Fast tiny wiggles; the atom barely excites and feels an AC Stark shift $\approx\Omega^2/4\delta$ .
Weak field plus damping	$\Omega\ll\Gamma\Rightarrow R_{\text{abs}}=\dfrac{\Omega^2/2}{\Gamma}\dfrac{(\Gamma/2)^2}{\delta^2+(\Gamma/2)^2}$	Coherences eliminated; dynamics collapse to incoherent rate equations, Lorentzian line.
Strong saturation	$\Omega\gg\Gamma\Rightarrow\rho_{ee}^{ss}\to\tfrac12$	Steady population saturates at one half; many cycles survive (Mollow regime).
Short pulse, no decay	$\Gamma t\ll1$	Coherent limit: only the pulse area $\theta=\int\Omega\,\mathrm{d}t$ matters; gates are area-defined.

Common trap"Off resonance you can still fully invert the atom if you just drive it long enough."

This is false, and the single most common mistake with the Rabi formula. The peak excited population is $\Omega^2/\tilde\Omega^2=\Omega^2/(\Omega^2+\delta^2)$ , a hard ceiling set by the detuning, not by how long you wait. A longer pulse only carries the state around the tilted cone again; because the cone never touches the excited pole, $P_e$ never exceeds $\Omega^2/\tilde\Omega^2$ . For full inversion you must sit on resonance, raise $\Omega$ until $\Omega\gg\delta$ , or sweep through resonance adiabatically. A subtler trap: some texts define $\Omega$ with the full field $E_0$ and others with $E_0/2$ per rotating component, so a factor of two can drift into $t_\pi$ — always check the field convention.

Worked examples

1 · Rabi frequency and π-pulse time for an optical transition (confidence builder)

A laser of intensity $I=10\ \mathrm{mW\,cm^{-2}}=100\ \mathrm{W\,m^{-2}}$ drives, on resonance, a transition of dipole matrix element $d=2.5\times10^{-29}\ \mathrm{C\,m}$ (a few $ea_0$ ). Find the Rabi frequency and the $\pi$ -pulse duration.

Step 1. Convert intensity to field amplitude. For a plane wave $I=\tfrac12 c\varepsilon_0 E_0^2$ , so $E_0=\sqrt{2I/(c\varepsilon_0)}=\sqrt{2(100)/[(2.998\times10^8)(8.854\times10^{-12})]}=\sqrt{7.53\times10^4}=\mathbf{274\ V\,m^{-1}}$ . A modest field delivers milliwatt-per-square-centimetre intensities.

Step 2. Apply the definition (equation 1). With the dipole aligned to the field, $\Omega=dE_0/\hbar=(2.5\times10^{-29})(274)/(1.055\times10^{-34})=6.51\times10^{7}\ \mathrm{rad\,s^{-1}}$ , i.e. $\Omega/2\pi=\mathbf{10.4\ MHz}$ . The megahertz range, comparable to a natural linewidth.

Step 3. The $\pi$ -pulse time is $t_\pi=\pi/\Omega=\pi/(6.51\times10^{7})=\mathbf{48\ ns}$ . Drive for 48 ns and the population is fully inverted; for 96 ns it is back in the ground state.

Step 4. Cross-check: for $\Gamma=2\pi\times6\ \mathrm{MHz}$ the saturation parameter is $s=2\Omega^2/\Gamma^2\approx6$ , so the beam runs at about $6\,I_{\text{sat}}$ . Ω/2π = 10.4 MHz and t_π = 48 ns.

2 · Off-resonant flopping and the light shift

The same atom is now detuned. (a) With $\Omega/2\pi=10\ \mathrm{MHz}$ and $\delta=\Omega$ , find the generalized Rabi frequency and maximum excited population. (b) With the same $\Omega$ but a far detuning $\delta/2\pi=1\ \mathrm{GHz}$ , find the AC Stark shift.

Step 1. With $\delta=\Omega$ , equation (2) gives $\tilde\Omega=\sqrt{\Omega^2+\Omega^2}=\sqrt2\,\Omega$ , so $\tilde\Omega/2\pi=\sqrt2\times10=\mathbf{14.1\ MHz}$ — about 40% faster than on resonance.

Step 2. Peak population: $P_e^{\max}=\Omega^2/\tilde\Omega^2=\Omega^2/(2\Omega^2)=\mathbf{0.50}$ . Even a perfectly timed pulse reaches only half inversion — a detuning equal to $\Omega$ already halves it.

Step 3. Far-detuned light shift: with $\delta/2\pi=1\ \mathrm{GHz}\gg\Omega$ , equation (5) gives $\Delta_{\text{Stark}}=\Omega^2/(4\delta)=(2\pi\times10^7)^2/[4(2\pi\times10^9)]=2\pi\times2.5\times10^{4}\ \mathrm{rad\,s^{-1}}$ , i.e. $\Delta_{\text{Stark}}/2\pi=\mathbf{25\ kHz}$ . The atom is essentially unexcited ( $P_e^{\max}=\Omega^2/\delta^2\approx10^{-4}$ ) but its level is pushed 25 kHz — the principle of an optical dipole trap. Ω̃/2π = 14.1 MHz, P_e^max = 0.50, light shift = 25 kHz.

Physics of the portrait

The Rabi frequency carries the name of the physicist who first solved the two-level problem in a rotating field and built a measurement method on it.

Origins · 1 minIsidor Isaac Rabi and how this was discovered

Isidor Isaac Rabi

1898–1988 · molecular-beam resonance and the driven two-level atom

Isidor Isaac Rabi, born in Galicia and raised in New York, spent the 1930s at Columbia making atoms and nuclei report their own internal frequencies. In a 1937 paper on "space quantization in a gyrating magnetic field" he solved exactly the problem that now bears his name: a two-level system driven by an oscillating field flops between its states at a rate fixed by the field strength and the detuning — the Rabi frequency and its detuned generalization. By 1938 his group had turned this into the molecular-beam magnetic-resonance method, in which a beam is deflected only when the drive matches an internal transition, reading off a nuclear moment with unprecedented accuracy. This won the 1944 Nobel Prize and seeded a dynasty — his student Norman Ramsey extended it with the separated-oscillatory-field method behind every atomic clock, and the same flopping reappeared in NMR, the Autler–Townes effect, and eventually every qubit. A classical field pushing a quantum two-level system at $\Omega$ is Rabi's enduring picture of light meeting matter.

History

1937

— Rabi solves the two-level system in a gyrating magnetic field, giving the flopping formula

1938

— Rabi's molecular-beam magnetic-resonance method measures nuclear moments

1944

— Rabi receives the Nobel Prize in Physics for the resonance method

1949

— Ramsey introduces the separated-oscillatory-field method of two π/2-pulses

1955

— Autler and Townes observe the dynamic Stark splitting of a driven level

1969

— Mollow predicts the three-peak fluorescence spectrum of a strongly driven atom

Going further

Texts. Foot, Atomic Physics, Oxford (2005), Ch. 7 · Loudon, The Quantum Theory of Light, 3rd ed., Oxford (2000), Ch. 2 · Saleh and Teich, Fundamentals of Photonics, 2nd ed., Wiley (2007), Ch. 13 · Demtröder, Laser Spectroscopy, 4th ed., Springer (2008) · Allen and Eberly, Optical Resonance and Two-Level Atoms, Dover (1987) · Steck, Rubidium 87 D Line Data. Key results to keep in mind: the RWA Hamiltonian eigenvalue gap is $\hbar\sqrt{\Omega^2+\delta^2}$ ; the Rabi formula is $P_e=(\Omega^2/\tilde\Omega^2)\sin^2(\tilde\Omega t/2)$ ; the steady-state excited population is $\rho_{ee}^{ss}=(\Omega^2/4)/(\delta^2+\Gamma^2/4+\Omega^2/2)$ with a $1/2$ ceiling. Worked-example anchor numbers: $E_0=274\ \mathrm{V\,m^{-1}}$ at $100\ \mathrm{W\,m^{-2}}$ , $\Omega/2\pi=10.4\ \mathrm{MHz}$ for $d=2.5\times10^{-29}\ \mathrm{C\,m}$ , $t_\pi=48\ \mathrm{ns}$ ; $\tilde\Omega=\sqrt2\,\Omega$ and $P_e^{\max}=0.5$ at $\delta=\Omega$ ; light shift $25\ \mathrm{kHz}$ at $\delta/2\pi=1\ \mathrm{GHz}$ .

Prerequisites. Transition Dipole Moment · Electric Dipole Approximation · Stimulated Emission. Leads to. Radiative Lifetime and Natural Linewidth · Lorentzian Line Shape · Resonant Absorption.

Ask the tutor: Want to walk the Bloch vector through a detuned pulse and watch where the population tops out — or see how a Ramsey pair of π/2-pulses turns detuning into fringes?

Is the Rabi frequency the frequency of the laser light?

No — this is the most common conflation. The optical frequency $\omega$ is hundreds of terahertz; the Rabi frequency $\Omega$ is the far slower rate at which the population sloshes between the levels, typically kilohertz to megahertz — $\omega$ is the fast carrier, $\Omega$ the slow envelope. Formally $\Omega=-\mathbf{d}_{eg}\cdot\mathbf{E}_0/\hbar$ depends on the field amplitude, not its frequency; the laser frequency enters only through the detuning $\delta=\omega-\omega_0$ . Turn the intensity up and $\Omega$ grows; retune and only $\delta$ changes.

Why does the population oscillate at Ω̃ but the amplitude beat at Ω̃/2?

Because the excited-state amplitude is a sine, $c_e\propto\sin(\tilde\Omega t/2)$ , while the probability is its square, $P_e\propto\sin^2(\tilde\Omega t/2)$ . Squaring doubles the frequency ( $\sin^2x=\tfrac12(1-\cos2x)$ ), so the probability oscillates at $\tilde\Omega$ even though the amplitude carries $\tilde\Omega/2$ . This is why the population period is $2\pi/\tilde\Omega$ and a $\pi$ -pulse — a half cycle — has area only $\pi$ , not $2\pi$ . Keeping straight which quantity you time removes most factor-of-two confusion.

What is the pulse area, and why does only the area matter for a gate?

The pulse area is $\theta=\int_0^t\Omega(t')\,\mathrm{d}t'$ , the time integral of the instantaneous Rabi frequency. On resonance it is exactly the Bloch-vector rotation angle, so the final state depends only on this integral — not on whether the pulse was short and strong or long and weak, square or shaped. That lets a controller absorb fluctuating laser power by adjusting duration to keep the area at $\pi$ or $\pi/2$ . The rule holds only while decay is negligible ( $\Gamma t\ll1$ ) and on resonance; off resonance the axis tilts and you need the full generalized-Rabi geometry.

When does coherent flopping turn into rate-equation absorption?

When dephasing or decay outpaces the drive, $\Omega\ll\Gamma$ . The coherence then relaxes almost as fast as it is created and can be adiabatically eliminated: the population stops oscillating and climbs with an absorption rate $R=(\Omega^2/2)(\Gamma/2)/[\delta^2+(\Gamma/2)^2]$ , a Lorentzian in detuning. This is precisely the incoherent Einstein picture — exponential approach to steady state, no flopping — recovered as the strong-damping limit of the optical Bloch equations. The Rabi frequency thus interpolates: coherent oscillation for $\Omega\gg\Gamma$ , rate equations for $\Omega\ll\Gamma$ .

How is the Rabi frequency related to laser intensity and saturation?

Since $\Omega=dE_0/\hbar$ and $I=\tfrac12c\varepsilon_0E_0^2$ , the Rabi frequency scales as $\Omega\propto\sqrt I$ — double the intensity and $\Omega$ rises by $\sqrt2$ . The clean packaging is the on-resonance saturation parameter $s=2\Omega^2/\Gamma^2=I/I_{\text{sat}}$ , which defines $I_{\text{sat}}$ as the intensity where $s=1$ , i.e. $\Omega=\Gamma/\sqrt2$ . For the rubidium-87 cooling line ( $\Gamma=2\pi\times6.07\ \mathrm{MHz}$ , $I_{\text{sat}}\approx1.67\ \mathrm{mW\,cm^{-2}}$ ) that gives $\Omega/2\pi\approx4.3\ \mathrm{MHz}$ at one saturation intensity.

Why can a detuned drive never fully invert the atom?

Geometrically, detuning tilts the effective Bloch vector out of the equator by $\arctan(\delta/\Omega)$ , so the state precesses on a cone about that tilted axis — and a tilted cone cannot reach the excited pole unless the axis lies flat in the equator, which happens only at $\delta=0$ . Algebraically, $P_e^{\max}=\Omega^2/\tilde\Omega^2=\Omega^2/(\Omega^2+\delta^2)<1$ for any $\delta\neq0$ . Waiting longer only sends the state around the cone again. For full inversion, raise $\Omega$ until it dominates $\delta$ , tune onto resonance, or sweep adiabatically through resonance.