Semiclassical Two-Level Atom

Truncate an atom to its two most relevant states and shine a classical light wave on it: that single act of amputation is the most productive idealization in atomic physics. Every real atom has infinitely many levels, but when a laser sits near one resonance it couples one pair — a ground state $|g\rangle$ and an excited state $|e\rangle$ separated by the Bohr frequency $\omega_0=(E_e-E_g)/\hbar$ — far more strongly than any other, and the rest of the spectrum is quietly discarded. What survives is a quantum coin driven by a classical field through the electric-dipole coupling $-\mathbf{d}\cdot\mathbf{E}(t)$ of aq-optics-electric-dipole-approximation, its strength set by the transition dipole moment $\mathbf{d}_{eg}$ of aq-optics-transition-dipole-moment. The payoff is that near-resonant light–matter interaction collapses onto two numbers: the Rabi frequency $\Omega$ , how hard the field grips the dipole, and the detuning $\Delta=\omega-\omega_0$ , how far the drive sits from resonance. In the rotating-wave approximation these govern everything: an atom driven on resonance sloshes coherently between its levels — Rabi flopping — a purely quantum effect in which a strong enough pulse fully inverts the population and back again. That coherent picture exposes the model's one honest failure: a prescribed classical field can only stimulate, never spontaneously emit, so the norm is conserved and the oscillations would ring forever. Spontaneous decay must be reinstated by hand as a rate $\Gamma$ — the inverse radiative lifetime, fixed ultimately by the Einstein $A$ coefficient and QED. Adding it turns the Schrödinger picture into the optical Bloch equations, whose steady state is the saturated, power-broadened Lorentzian line underlying laser cooling, atomic clocks, and every spectroscopy experiment on Earth.

Key takeaways

The atom is kept quantum (two eigenstates $|g\rangle,|e\rangle$ ) while the field is a prescribed classical wave $\mathbf{E}_0\hat{\boldsymbol\epsilon}\cos\omega t$ ; they couple through the dipole interaction $-\mathbf{d}\cdot\mathbf{E}(t)$ .
In the rotating-wave approximation the dynamics reduce to two parameters: the Rabi frequency $\Omega=-\mathbf{d}_{eg}\cdot\hat{\boldsymbol\epsilon}\,E_0/\hbar$ (coherent exchange rate) and the detuning $\Delta=\omega-\omega_0$ .
On resonance the population floops at $\Omega$ and fully inverts at a $\pi$ -pulse ( $\Omega t=\pi$ ); off resonance it oscillates faster, at $\Omega'=\sqrt{\Omega^2+\Delta^2}$ , but never fully inverts.
Spontaneous emission cannot appear in a classical field, so it is added by hand as a decay rate $\Gamma$ , turning the coherent flopping into the damped optical Bloch equations and a saturated Lorentzian line.
The model reproduces absorption, stimulated emission, saturation, power broadening, and the AC Stark shift, and recovers the Einstein $B$ coefficient — the bridge from rate equations to the quantized Jaynes–Cummings model.
Key idea: truncate the atom to two levels and drive it with a classical field through $-\mathbf{d}\cdot\mathbf{E}$ ; in the rotating-wave approximation the entire physics collapses onto two numbers, the Rabi frequency $\Omega$ (coupling) and the detuning $\Delta$ (off-resonance), producing coherent Rabi oscillations that spontaneous emission — added by hand as $\Gamma$ — damps into a saturated Lorentzian line.

The core insight

Truncate the atom to two levels and drive it with a classical field through $-\mathbf{d}\cdot\mathbf{E}$ ; in the rotating-wave approximation the entire physics collapses onto two numbers, the Rabi frequency $\Omega$ (coupling strength) and the detuning $\Delta$ (off-resonance), producing coherent Rabi oscillations that spontaneous emission — added by hand as $\Gamma$ — damps into a saturated Lorentzian line.

Where this lives today

The semiclassical two-level atom is not a historical stepping stone but the daily working model of the cold-atom and quantum-information enterprise. Every optical atomic clock — now defining the second to better than one part in $10^{18}$ — is a two-level atom interrogated by a laser, its transition probability read out through exactly the flopping and Ramsey physics derived above; the clock's stability demands a low saturation parameter so power broadening does not wash out the natural $\Gamma$ . In trapped-ion and neutral-atom quantum computers the $\pi$ - and $\pi/2$ -pulses of Worked Example 3 are the literal single-qubit gates, with the coherence budget set by $\Gamma$ versus the gate rate $\Omega$ . The same equations run the machines that make cold atoms. A magneto-optical trap and a Zeeman slower are radiation-pressure devices whose force is $\hbar k\,R_{\mathrm{sc}}$ — the $9\times10^{5}\ \mathrm{m/s^2}$ of Worked Example 1 — while optical dipole traps and tweezers use the far-detuned AC Stark shift $\delta E=\hbar\Omega^2/4\Delta$ as a conservative potential, choosing $|\Delta|\gg\Gamma$ so real scattering is suppressed. Push the model one step further — replace $E_0\cos\omega t$ by a single quantized cavity mode — and the classical Rabi frequency becomes the vacuum Rabi frequency, spontaneous emission emerges on its own, and the model becomes the Jaynes–Cummings Hamiltonian of cavity and circuit QED.

The physical picture

Picture the atom as a single quantum degree of freedom with only two accessible rungs. A classical light wave grabs the atomic dipole and rocks the amplitude between the rungs at a rate set by how hard it pushes — the field amplitude times the transition dipole moment. Unlike a classical charge that would simply absorb energy and heat up, the quantum two-level system coherently sloshes population: it is a probability amplitude, not a temperature, that the field drives. A strong enough resonant pulse flips the atom entirely into the excited state and, if the drive continues, back to the ground state — a clean sine-squared oscillation. Resonance is a phase-matching condition. When the drive frequency equals the Bohr frequency, every optical cycle nudges the amplitude in the same sense and the flopping is complete, cresting at unity. Detune the drive and successive pushes fall out of step: the atom oscillates faster, at $\Omega'=\sqrt{\Omega^2+\Delta^2}$ , but with reduced amplitude, its peak capped at $\Omega^2/(\Omega^2+\Delta^2)$ and never reaching unity. The width in $\Delta$ over which the atom responds strongly is the power-broadened linewidth.

Figure

Rabi flopping of the excited-state probability. On resonance Pe = sin squared of Ωt over 2 climbs to unity at the π-pulse Ωt = π, then returns to zero at Ωt = 2π. With detuning Δ = Ω the oscillation speeds up to Ω′ = √2 Ω and its peak falls to only 0.5 — a detuned drive can never fully invert the atom

The model has a built-in blind spot, and it is the honest heart of the subject. A prescribed classical field can only stimulate transitions; it cannot spontaneously eject the atom's stored energy into empty modes, because those modes are absent from the theory. Spontaneous emission is a vacuum-fluctuation effect that lives only in the quantized field, so in the semiclassical picture we bolt it on as an exponential decay at rate $\Gamma$ , which damps the flopping until drive and relaxation reach a standoff — the saturated steady state. As the intensity climbs, that steady state does two things at once: it saturates toward equal populations ( $\rho_{ee}\to\tfrac12$ ) and it broadens, because a stronger field forces the atom to respond even when badly detuned.

Figure

Steady-state excited fraction ρₑₑ versus detuning Δ over Γ. As the on-resonance saturation s₀ grows from 0.2 to 2 to 20, the Lorentzian both climbs toward the saturation ceiling ρₑₑ = 0.5 and broadens: its full width at half maximum is Γ times √(1 + s₀), reaching about 4.6 Γ at s₀ = 20 — power broadening

The core equations

The starting point is the full two-level Hamiltonian: the bare excited-state energy plus the dipole coupling to the classical field. With the ground energy set to zero the internal part is diagonal and the interaction entirely off-diagonal.

H = \hbar\omega_0\,|e\rangle\langle e| \;-\; \mathbf{d}\cdot\mathbf{E}(t) \qquad (1)

where: $\omega_0=(E_e-E_g)/\hbar$ — atomic Bohr frequency; $\mathbf{d}=-e\mathbf{r}$ — electric-dipole operator; $\mathbf{E}(t)=E_0\hat{\boldsymbol\epsilon}\cos\omega t$ — classical monochromatic drive of amplitude $E_0$ , polarization $\hat{\boldsymbol\epsilon}$ . By parity only the off-diagonal transition matrix element of $\mathbf{d}$ survives, so the field couples the levels but shifts neither on its own. The entire physics is carried by two numbers built from the field and the transition dipole: the Rabi frequency $\Omega$ , the field amplitude times the projected dipole over $\hbar$ , and the detuning $\Delta$ , how far the laser sits from resonance.

\Omega \equiv -\frac{\mathbf{d}_{eg}\cdot\hat{\boldsymbol\epsilon}\,E_0}{\hbar},\qquad \Delta \equiv \omega-\omega_0 \qquad (2)

where: $\mathbf{d}_{eg}=\langle e|\mathbf{d}|g\rangle$ — transition dipole moment (off-diagonal); $\Omega$ — Rabi frequency, the coherent exchange rate, $\propto E_0\propto\sqrt{I}$ ; $\Delta$ — detuning, positive for blue drive. $\Omega$ says how hard the field grips the dipole, $\Delta$ how far off resonance it sits — every result below is a function of these two. Expanding $\cos\omega t$ into $e^{\pm i\omega t}$ and moving to the interaction picture splits the coupling into a slow co-rotating part ( $\sim\Delta$ ) and a fast counter-rotating part ( $\sim\omega+\omega_0$ ). Dropping the fast part — the rotating-wave approximation, valid when $|\Delta|\ll\omega_0$ — leaves two coupled amplitude equations.

i\dot c_e = -\tfrac{\Omega}{2}e^{-i\Delta t}c_g,\qquad i\dot c_g = -\tfrac{\Omega^*}{2}e^{+i\Delta t}c_e \qquad (3)

where: $c_g,c_e$ — probability amplitudes with $|c_g|^2+|c_e|^2=1$ . Each amplitude drives the other at rate $\Omega/2$ , phased by the detuning — the seed of coherent oscillation. Solving (3) with the atom initially in the ground state gives the central result of the coherent theory: Rabi flopping. The excited-state probability oscillates at the generalized Rabi frequency, with an amplitude that is unity only on resonance.

P_e(t) = \frac{\Omega^2}{\Omega'^2}\,\sin^2\!\left(\frac{\Omega' t}{2}\right),\qquad \Omega'=\sqrt{\Omega^2+\Delta^2} \qquad (4)

where: $P_e(t)$ — probability of finding the atom excited at time $t$ ; $\Omega'$ — generalized (off-resonant) Rabi frequency. The prefactor $\Omega^2/\Omega'^2$ caps the peak excitation below one whenever $\Delta\neq0$ , while $\Omega'$ speeds the oscillation — detuning trades depth for pace. To include spontaneous decay we promote the state to a density matrix and add a Lindblad relaxation term by hand. This is the optical Bloch equation: coherent Hamiltonian evolution plus dissipation the classical field cannot supply.

\dot\rho = -\frac{i}{\hbar}[H,\rho] + \mathcal{L}_\Gamma[\rho] \qquad (5)

where: $\rho$ — $2\times2$ density matrix, diagonal the populations, off-diagonal $\rho_{eg}$ the coherence; $\mathcal{L}_\Gamma$ — relaxation decaying the excited population at $\Gamma$ and the coherence at $\Gamma/2$ . Here $\Gamma=1/\tau$ is the inverse radiative lifetime, imported from the Einstein $A$ coefficient because semiclassics cannot generate it. Setting all time derivatives to zero gives the steady-state excited population — a Lorentzian in detuning, saturating and broadening as the drive strengthens.

\rho_{ee}^{\mathrm{ss}} = \frac{\Omega^2/4}{\Delta^2+\Gamma^2/4+\Omega^2/2},\qquad s = \frac{\Omega^2/2}{\Delta^2+\Gamma^2/4},\qquad R_{\mathrm{sc}}=\frac{\Gamma}{2}\,\frac{s}{1+s} \qquad (6)

where: $\rho_{ee}^{\mathrm{ss}}=\tfrac12\,s/(1+s)$ — steady-state excited probability; $s$ — saturation parameter, $=I/I_{\mathrm{sat}}$ on resonance; $R_{\mathrm{sc}}$ — photon scattering rate, saturating at $\Gamma/2$ . At low drive $\rho_{ee}\propto\Omega^2\propto I$ (linear absorption); at high drive the populations equalize and scattering caps out. The width of that Lorentzian is not fixed at the natural value $\Gamma$ : a strong drive broadens it. This is power broadening, the reason narrow-line spectroscopy demands low saturation.

\Gamma_{\mathrm{FWHM}} = \Gamma\sqrt{1+\Omega^2/(\Gamma^2/2)} = \Gamma\sqrt{1+s_0} \qquad (7)

where: $\Gamma_{\mathrm{FWHM}}$ — full width at half maximum of the saturated line; $s_0=2\Omega^2/\Gamma^2$ — on-resonance saturation parameter. The natural width $\Gamma$ is recovered only as $\Omega\to0$ ; every real measurement broadens by $\sqrt{1+s_0}$ .

Derivation ARabi oscillations from the rotating-wave equations

The goal is to solve the two coupled rotating-wave amplitude equations exactly and extract the flopping formula $P_e(t)=(\Omega^2/\Omega'^2)\sin^2(\Omega't/2)$ , showing how detuning both speeds the oscillation and caps its amplitude below unity. We start from equations (3) with the atom in the ground state, $c_g(0)=1$ , $c_e(0)=0$ , remove the explicit time dependence, decouple into a single constant-coefficient equation, and square the amplitude.

Step 1. Absorb the phase by defining $\tilde c_e=c_e\,e^{i\Delta t}$ , giving the constant-coefficient system $i\dot{\tilde c}_e=\Delta\tilde c_e-\tfrac{\Omega}{2}c_g$ , $i\dot c_g=-\tfrac{\Omega^*}{2}\tilde c_e$ . Why: a frame co-rotating with the drive freezes the fast optical phase, leaving the slow beat at the detuning $\Delta$ — the quantity the atom actually feels.

Step 2. Differentiate and substitute to eliminate $c_g$ , giving $\ddot{\tilde c}_e+i\Delta\,\dot{\tilde c}_e+\tfrac{|\Omega|^2}{4}\tilde c_e=0$ . Why: its characteristic roots $(-\Delta\pm\Omega')/2$ define the generalized Rabi frequency $\Omega'=\sqrt{\Omega^2+\Delta^2}$ that governs the whole motion.

Step 3. Impose the initial conditions to get $\tilde c_e(t)=i(\Omega/\Omega')\sin(\Omega' t/2)\,e^{-i\Delta t/2}$ , and take the modulus squared to reach $P_e(t)=(\Omega^2/\Omega'^2)\sin^2(\Omega't/2)$ . Why: the prefactor $\Omega^2/\Omega'^2$ is the amplitude cap (unity only at $\Delta=0$ ) while the argument carries the faster off-resonant pace — one formula, both effects.

Step 4. Specialize to resonance $\Delta=0$ : $P_e(t)=\sin^2(\Omega t/2)$ , so the atom fully inverts at $\Omega t=\pi$ (a $\pi$ -pulse) and returns at $\Omega t=2\pi$ . Why: this is coherent control in its purest form — complete, reversible transfer, the operation behind qubit gates and Ramsey spectroscopy. *Verified: reduces to complete inversion at resonance and to vanishing amplitude as $|\Delta|\to\infty$ ; matches Foot, Atomic Physics Ch. 7 and Loudon Ch. 2.*

Go deeper · Derivation B · 1 minSteady state of the optical Bloch equations

Derivation BSteady state of the optical Bloch equations

The goal is to set every time derivative to zero in the driven, damped Bloch equations and solve for the steady-state excited population, exposing the saturated Lorentzian $\rho_{ee}^{\mathrm{ss}}=\tfrac12\,s/(1+s)$ and the power-broadened width $\Gamma\sqrt{1+s_0}$ . We work with the inversion $w=\rho_{ee}-\rho_{gg}$ and the slow coherence $\tilde\rho_{eg}$ , driven at $\Omega$ , detuned by $\Delta$ , relaxing at $\Gamma$ (population) and $\Gamma/2$ (coherence) inserted by hand.

Step 1. Set $\dot{\tilde\rho}_{eg}=0$ in $\dot{\tilde\rho}_{eg}=-(\Gamma/2+i\Delta)\tilde\rho_{eg}-i(\Omega/2)w$ to get the steady coherence $\tilde\rho_{eg}=\dfrac{-i\Omega/2}{\Gamma/2+i\Delta}\,w$ . Why: the drive builds a coherence in quadrature with the field, and its imaginary part feeds real population upward — the microscopic origin of absorption.

Step 2. Impose population balance $\Gamma\rho_{ee}=\Omega\,\mathrm{Im}(\tilde\rho_{eg})$ and substitute Step 1. Why: in steady state nothing accumulates, so every photon coherently absorbed must be spontaneously re-emitted — this balance pins the population.

Step 3. Use $\rho_{ee}+\rho_{gg}=1$ to solve, giving $\rho_{ee}^{\mathrm{ss}}=\dfrac{\Omega^2/4}{\Delta^2+\Gamma^2/4+\Omega^2/2}=\tfrac12\,\dfrac{s}{1+s}$ , $s=(\Omega^2/2)/(\Delta^2+\Gamma^2/4)$ . Why: the compact $\tfrac12 s/(1+s)$ makes both limits transparent — linear when weak, ceiling at $\tfrac12$ when strong — and defines $I_{\mathrm{sat}}$ through $s=1$ .

Step 4. The denominator as a function of $\Delta$ has half-width $\sqrt{\Gamma^2/4+\Omega^2/2}$ , so $\Gamma_{\mathrm{FWHM}}=\Gamma\sqrt{1+2\Omega^2/\Gamma^2}=\Gamma\sqrt{1+s_0}$ . Why: the drive adds its own term to the denominator, widening the resonance — power broadening — with only $\Gamma$ surviving as $\Omega\to0$ . *Verified: weak-field limit $\Omega\ll\Gamma$ gives a Lorentzian of width $\Gamma$ with $\rho_{ee}\propto\Omega^2$ (linear, Einstein $B$ ); strong-field limit gives $\rho_{ee}\to\tfrac12$ (saturation). Consistent with Saleh and Teich Ch. 13.*

Go deeper · Derivation C · 1 minRecovering the Einstein B coefficient

Derivation CRecovering the Einstein B coefficient

The goal is to show that the weak-field steady-state excitation rate of the semiclassical model is exactly the Einstein stimulated rate $B\,\rho(\omega_0)$ , so the phenomenological $\Gamma$ we inserted is not free but tied to $B$ — and hence the vacuum — by detailed balance. We take the perturbative limit, integrate the Lorentzian over detuning, re-express the field through the energy density, orientation-average the dipole, and identify the coefficient.

Step 1. For $\Omega\ll\Gamma$ the excitation rate is $W_{g\to e}=\Gamma\rho_{ee}^{\mathrm{ss}}\to\dfrac{\Omega^2}{4}\dfrac{\Gamma}{\Delta^2+\Gamma^2/4}$ ; integrating over $\Delta$ gives $\int W\,\mathrm{d}\Delta=\pi\Omega^2/2$ . Why: $\Gamma$ cancels from the integrated rate, leaving something set only by field strength — the fingerprint of a stimulated process.

Step 2. Substitute $\Omega^2=|d_{eg}|^2E_0^2/\hbar^2$ and $u=\tfrac12\varepsilon_0E_0^2$ (spectral density $\rho(\omega)$ for broadband light). Why: the Einstein picture is written in terms of radiation energy density, so the drive amplitude must be translated into $u$ first.

Step 3. Orientation-average $|\mathbf{d}_{eg}\cdot\hat{\boldsymbol\epsilon}|^2$ over an isotropic field (factor $\tfrac13$ ) and identify $B=\dfrac{\pi|d_{eg}|^2}{3\varepsilon_0\hbar^2}$ , so the rate is $B\,\rho(\omega_0)$ . Why: the $\tfrac13$ is the average of $\cos^2$ over the sphere — exactly the Einstein $B$ of the rate-equation treatment.

Step 4. Thermal equilibrium with the Planck spectrum forces $A/B=\hbar\omega_0^3/(\pi^2c^3)$ , locking $\Gamma=A$ to $B$ . Why: the vacuum we could not describe semiclassically fixes $\Gamma$ through equilibrium thermodynamics — and QED later derives the identical $A$ from first principles. *Verified: the extracted $B$ matches the standard Einstein-coefficient expression, and with the Planck detailed-balance relation reproduces $A=\Gamma$ , consistent with aq-optics-einstein-a-b-coefficients.*

Limiting cases

The two dimensionless ratios $\Omega/\Gamma$ (drive versus relaxation) and $\Delta/\Omega$ (detuning versus coupling) organize every regime of the model.

Case	Result	Meaning
Coherent limit ( $\Omega\gg\Gamma$ , resonant)	$P_e(t)=\sin^2(\Omega t/2)$	Undamped flopping; $\Omega t=\pi$ inverts, $2\pi$ returns. Decay negligible over a cycle — the qubit-gate regime.
Weak-field / rate-equation ( $\Omega\ll\Gamma$ )	$\rho_{ee}^{\mathrm{ss}}\approx\dfrac{\Omega^2/4}{\Delta^2+\Gamma^2/4}$	Excitation is a Lorentzian of half-width $\Gamma/2$ , linear in intensity — where the Einstein $B$ coefficient and rate equations emerge.
Strong saturation ( $\Omega\gg\Gamma$ , steady state)	$\rho_{ee}^{\mathrm{ss}}\to\tfrac12$	Populations equalize; the medium bleaches and scattering saturates at $\Gamma/2$ . A classical field cannot pump past inversion.
Large detuning ( $\|\Delta\|\gg\Omega,\Gamma$ )	$\delta E\approx\dfrac{\hbar\Omega^2}{4\Delta}$	Negligible real excitation, but a second-order light shift (AC Stark shift) — the conservative potential behind far-off-resonant dipole traps.

Common trapA classical field can make an excited atom decay on its own

It cannot. The semiclassical Schrödinger equation is unitary: it conserves the norm and predicts eternal, undamped Rabi oscillations, so an isolated excited atom in a purely classical field would ring forever and never emit. Spontaneous emission is a coupling to the quantized vacuum — it does not exist in a theory where the field is a prescribed classical function. That is why $\Gamma$ must be inserted by hand into the optical Bloch equations, its value borrowed from the Einstein $A$ coefficient or QED. A companion trap is to conflate the two rates: $\Omega$ is coherent, reversible, and $\propto\sqrt{I}$ , whereas $\Gamma$ is incoherent, irreversible, and independent of the drive — never substitute one for the other.

Worked examples

1 · How hard can light push a sodium atom (the concrete anchor)

A sodium atom is driven hard on its D2 line, $\lambda=589$ nm, natural linewidth $\Gamma/2\pi=9.79$ MHz ( $\tau=16.2$ ns). At full saturation the scattering rate maxes at $\Gamma/2$ . Find the maximum radiation-pressure acceleration and the distance to stop a thermal atom leaving a $600$ K oven at its most-probable speed — the physics of the Zeeman slower.

Step 1. Each scattered photon delivers one recoil $\hbar k$ with $k=2\pi/\lambda=1.067\times10^{7}\ \mathrm{m^{-1}}$ , so $\hbar k=1.12\times10^{-27}\ \mathrm{kg\,m/s}$ . For a sodium mass $m=3.82\times10^{-26}\ \mathrm{kg}$ the recoil velocity is $v_{\mathrm{rec}}=\hbar k/m=\mathbf{2.95\ cm/s}$ per photon.

Step 2. The scattering rate saturates at $R_{\mathrm{sc}}^{\max}=\Gamma/2$ . With $\Gamma=2\pi\times9.79\times10^{6}=6.15\times10^{7}\ \mathrm{s^{-1}}$ , $R_{\mathrm{sc}}^{\max}=3.08\times10^{7}\ \mathrm{s^{-1}}$ . The maximum acceleration is $a_{\max}=v_{\mathrm{rec}}\,R_{\mathrm{sc}}^{\max}=\mathbf{9.1\times10^{5}\ m/s^{2}}$ — about $\mathbf{9.2\times10^{4}\,g}$ .

Step 3. A sodium atom leaving a $600$ K oven has most-probable speed $v_p=\sqrt{2k_BT/m}=\mathbf{659\ m/s}$ . Stopping it takes $t=v_p/a_{\max}=\mathbf{0.73\ ms}$ over a distance $d=v_p^{2}/(2a_{\max})=\mathbf{0.24\ m}$ — the fraction-of-a-metre length of a real Zeeman slower, set by a single two-level saturation formula.

2 · Saturation and power broadening of rubidium-87

The rubidium-87 D2 cooling transition has $\Gamma/2\pi=6.07$ MHz. A probe drives it on resonance at three times its saturation intensity, $s_0=I/I_{\mathrm{sat}}=3$ . Find the steady-state excited fraction, the photon scattering rate, and the observed linewidth.

Step 1. The steady-state excited population is $\rho_{ee}^{\mathrm{ss}}=\tfrac12\,s_0/(1+s_0)=\tfrac12\times3/4=\mathbf{0.375}$ — still below the ceiling of $0.5$ , so even three times $I_{\mathrm{sat}}$ does not fully bleach the atom.

Step 2. The scattering rate is $R_{\mathrm{sc}}=(\Gamma/2)\,s_0/(1+s_0)$ . With $\Gamma=2\pi\times6.07\times10^{6}=3.81\times10^{7}\ \mathrm{s^{-1}}$ , $R_{\mathrm{sc}}=(1.91\times10^{7})\times0.75=\mathbf{1.43\times10^{7}\ photons/s}$ — one every $70$ ns, approaching but never exceeding the ceiling $\Gamma/2$ .

Step 3. The line is power-broadened to $\Gamma_{\mathrm{FWHM}}=\Gamma\sqrt{1+s_0}=2\Gamma$ , so $\Gamma_{\mathrm{FWHM}}/2\pi=\mathbf{12.1\ MHz}$ — twice the natural width. Measuring the true $6.07$ MHz linewidth requires $s_0\ll1$ , where $\sqrt{1+s_0}\to1$ .

3 · Designing a π-pulse and the cost of detuning

A resonant drive gives a Rabi frequency $\Omega/2\pi=1$ MHz. How long a pulse fully inverts the atom, and how long makes an equal superposition? If the laser is instead detuned by $\Delta=\Omega$ , what is the largest excitation any pulse can now reach, and how fast does the population oscillate?

Step 1. On resonance $P_e=\sin^2(\Omega t/2)$ , so full inversion needs $\Omega t=\pi$ , i.e. $t_\pi=\pi/\Omega=\mathbf{500\ ns}$ . An equal superposition (a $\pi/2$ -pulse) needs half that, $t_{\pi/2}=\mathbf{250\ ns}$ — the elementary operations of a qubit gate and of Ramsey interferometry.

Step 2. Detune to $\Delta=\Omega$ . The peak is now capped at $P_e^{\max}=\Omega^2/(\Omega^2+\Delta^2)=1/2$ , so no pulse of any length can excite the atom beyond 50% — complete inversion is impossible off resonance.

Step 3. The oscillation also speeds up: $\Omega'=\sqrt{\Omega^2+\Delta^2}=\sqrt{2}\,\Omega$ , so $\Omega'/2\pi=\mathbf{1.41\ MHz}$ and the peak arrives after $\Omega' t=\pi$ , i.e. $t=\mathbf{354\ ns}$ . Detuning buys speed at the price of depth — the fundamental trade of coherent control.

Physics of the portrait

The two-level atom is inseparable from the man who first solved it. Long before lasers, Isidor Isaac Rabi worked out how a two-state quantum system responds to an oscillating drive — for him a nuclear spin in a rotating magnetic field — and the flopping, the frequency, and the resonance all still bear his name.

Origins · 1 minIsidor Isaac Rabi and how this was discovered

Isidor Isaac Rabi

1898–1988 · Nobel 1944; molecular-beam magnetic resonance and the driven two-level atom

Isidor Isaac Rabi was born in Rymanów in Austrian Galicia in 1898 and carried to New York as an infant; he spent his scientific life at Columbia, where he built the molecular-beam magnetic-resonance method that earned him the 1944 Nobel Prize. In a 1937 paper on "space quantization in a gyrating magnetic field" he solved the problem at the heart of this article — two quantum states driven by a periodic field — and derived the transition probability we now call Rabi flopping, with its generalized frequency $\Omega'=\sqrt{\Omega^2+\Delta^2}$ . The next year his group demonstrated magnetic resonance in a molecular beam, measuring nuclear moments with unprecedented precision and inventing, in effect, the resonance spectroscopy behind NMR and MRI. His insight was that a two-state system driven on resonance does not merely absorb: it coherently and reversibly cycles, and the sharpness of that cycling in frequency is a ruler of exquisite accuracy. He suggested using such a resonance as a frequency standard — seeding the atomic clock — and his Columbia laboratory trained a remarkable lineage (Ramsey, Kusch, Lamb, Schwinger). When a modern experimenter speaks of a "π-pulse" or reads a Rabi frequency off an oscilloscope, they are using mathematics Rabi wrote down for a spinning nucleus nearly a century ago.

History

1917

— Einstein introduces the A and B coefficients, framing stimulated and spontaneous emission

1937

— Rabi solves the driven two-level spin problem and derives the flopping formula

1938

— Rabi, Zacharias, Millman and Kusch demonstrate molecular-beam magnetic resonance

1940

— Bloch and Siegert calculate the shift from the counter-rotating terms

1946

— Bloch writes the phenomenological Bloch equations for nuclear induction

1957

— Feynman, Vernon and Hellwarth recast any two-level system as a Bloch vector

1963

— Jaynes and Cummings quantize the field, replacing the classical drive

Going further

Texts. Foot, Atomic Physics (2005), Ch. 7 (Rabi flopping, optical Bloch equations) · Loudon, The Quantum Theory of Light, 3rd ed., Ch. 2 · Saleh and Teich, Fundamentals of Photonics, 2nd ed., Ch. 13 · Demtröder, Laser Spectroscopy, Vol. 1 · Cohen-Tannoudji, Dupont-Roc and Grynberg, Atom–Photon Interactions (the definitive dressed-state and optical-Bloch treatment) · Metcalf and van der Straten, Laser Cooling and Trapping (saturation, scattering force).

Provenance. The Rabi solution and rotating-wave approximation follow Foot Ch. 7 and Loudon Ch. 2 (reproduce $P_e=(\Omega^2/\Omega'^2)\sin^2(\Omega't/2)$ and the resonant $\sin^2(\Omega t/2)$ limit). The optical-Bloch steady state and saturation ( $\rho_{ee}^{\mathrm{ss}}=\tfrac12\,s/(1+s)$ , $R_{\mathrm{sc}}=(\Gamma/2)\,s/(1+s)$ , width $\Gamma\sqrt{1+s_0}$ ) follow Saleh and Teich Ch. 13. The $B$ -coefficient recovery and $A/B=\hbar\omega_0^3/(\pi^2c^3)$ should be checked against aq-optics-einstein-a-b-coefficients and Planck detailed balance. Worked numbers use tabulated data: sodium D2 ( $589$ nm, $\Gamma/2\pi=9.79$ MHz, $\tau=16.2$ ns) and rubidium-87 D2 ( $\Gamma/2\pi=6.07$ MHz); $a_{\max}=\hbar k\Gamma/2m\approx9\times10^5\ \mathrm{m/s^2}$ matches the Zeeman-slower value.

Builds on: Electric Dipole Approximation · Transition Dipole Moment · Selection Rules for Dipole Transitions · Stimulated Emission. Leads to: Radiative Lifetime and Natural Linewidth · Lorentzian Line Shape · Einstein A and B Coefficients · Resonant Absorption.

Ask the tutor: Want to see how a detuned drive traces a cone on the Bloch sphere, or how the vacuum Rabi frequency of cavity QED grows out of this same $\Omega$ ? Ask and we will build it step by step.

Why exactly two levels — real atoms have infinitely many?

Because near a resonance one pair of levels dominates so overwhelmingly that the rest can be ignored. The dipole coupling to any other level is suppressed by its large detuning: a level $1$ GHz away responds a thousand times more weakly than one on resonance a MHz away. As long as the drive is near a single transition and not accidentally resonant with a second, the other levels contribute only tiny, calculable corrections (small AC Stark shifts). The truncation fails when two transitions overlap — degenerate sublevels, closely spaced fine or hyperfine structure — where one keeps three or more levels, but the two-level atom remains the conceptual core.

What is the rotating-wave approximation actually throwing away?

When you expand $\cos\omega t$ into $e^{i\omega t}+e^{-i\omega t}$ , the coupling splits into a co-rotating term beating slowly at $\Delta=\omega-\omega_0$ and a counter-rotating term oscillating fast at $\omega+\omega_0$ . The rotating-wave approximation drops the fast term, which averages to nearly zero over any timescale the atom cares about — an excellent approximation whenever $|\Delta|\ll\omega_0$ and $\Omega\ll\omega_0$ , essentially always in optics. What it discards is a small level shift, the Bloch–Siegert shift of order $\Omega^2/\omega_0$ , negligible optically but measurable in strongly driven radio-frequency and circuit-QED systems.

How is the Rabi frequency related to the laser intensity?

Through the transition dipole moment. Since $\Omega=-\mathbf{d}_{eg}\cdot\hat{\boldsymbol\epsilon}\,E_0/\hbar$ and $I=\tfrac12\varepsilon_0 c\,E_0^2$ , the Rabi frequency scales as $\Omega\propto E_0\propto\sqrt{I}$ — double the intensity and $\Omega$ rises only by $\sqrt2$ . It is cleanest to express $\Omega$ through the saturation parameter: on resonance $s_0=2\Omega^2/\Gamma^2=I/I_{\mathrm{sat}}$ , so $\Omega=\Gamma\sqrt{I/2I_{\mathrm{sat}}}$ . This ties the coupling to a measurable intensity and to the tabulated saturation intensity (for example $1.67\ \mathrm{mW/cm^2}$ for the rubidium-87 cooling line), which is why experimentalists quote intensities in units of $I_{\mathrm{sat}}$ .

Why does the scattering rate saturate at Γ/2 and not at Γ?

Because a two-level atom holds at most half its population in the excited state under steady classical driving. Since $R_{\mathrm{sc}}=\Gamma\,\rho_{ee}^{\mathrm{ss}}$ and the populations equalize at strong drive ( $\rho_{ee}^{\mathrm{ss}}\to\tfrac12$ ), $R_{\mathrm{sc}}\to\Gamma/2$ . Physically, once excited the atom must wait to decay before it can be re-excited; at full saturation it spends equal time up and down, radiating at half the natural rate. No classical field can push past this ceiling — the only way to scatter faster is a shorter-lived transition (larger $\Gamma$ ), which is why cooling lines are chosen for their large linewidths.

What is the difference between the Rabi frequency Ω and the decay rate Γ?

They are opposite in almost every respect. $\Omega$ is coherent, reversible, and set by the drive: the rate at which the field cycles population up and down, proportional to $E_0$ (so to $\sqrt{I}$ ), and turning the laser off freezes it. $\Gamma$ is incoherent, irreversible, and intrinsic: the rate at which the excited state decays into the vacuum, fixed by the dipole moment and transition frequency through the Einstein $A$ coefficient, independent of the drive. Their ratio $\Omega/\Gamma$ decides whether you see clean oscillations ( $\Omega\gg\Gamma$ ) or an overdamped approach to a saturated line ( $\Omega\ll\Gamma$ ).

If the model cannot produce spontaneous emission, how does it give the natural linewidth?

It does not produce the linewidth from scratch — it imports it. The weak-field steady state is a Lorentzian whose half-width is $\Gamma/2$ , so the shape of the natural line is a genuine prediction of the optical Bloch equations, but the value of $\Gamma$ is put in by hand from the Einstein $A$ coefficient or, ultimately, from QED, where spontaneous emission arises from coupling to the vacuum. The model then correctly tells you how that width appears in a spectrum, how it broadens under power ( $\Gamma\sqrt{1+s_0}$ ), and how it feeds the scattering rate — it just cannot compute the bare $\Gamma$ .