13.1.1 Coherent population cycling versus incoherent rate saturation
A classical rate equation for a two-level atom driven by thermal or broadband radiation predicts that the excited-state population \(P_1(t)\) approaches a steady saturation value never exceeding \(1/2\) in the absence of population inversion. A coherent monochromatic drive near the Bohr transition frequency \(\omega_0 = (E_1 - E_0)/\hbar\) produces a qualitatively different signature: the population oscillates periodically between \(|0\rangle\) and \(|1\rangle\) as a function of pulse duration \(t\) or drive amplitude. Observing full-contrast Rabi oscillations that carry \(P_1(t)\) above \(1/2\) all the way toward \(1\) (and back down to \(0\)) demonstrates coherent unitary rotation of the quantum state vector rather than incoherent stochastic jumping between energy levels.
13.1.2 Separating state preparation, drive calibration, and readout fidelity
A physical Rabi experiment consists of three distinct operational stages: optical pumping or thermal relaxation that initializes the system near \(|0\rangle\), a gated radio-frequency, microwave, or laser pulse of duration \(t\), angular frequency \(\omega\), and Rabi rate \(\Omega\), and a projective readout stage such as state-selective fluorescence or dispersive cavity transmission. When an experimental trace fails to reach \(P_1 = 1\), the shortfall can arise from off-resonant detuning \(\Delta = \omega - \omega_0\), from longitudinal and transverse relaxation (\(T_1, T_2\)) during the pulse, or from state-preparation and measurement (SPAM) errors. Separating coherent control errors from SPAM offsets requires scanning both pulse duration \(t\) and detuning \(\Delta\) (a chevron scan).
13.1.3 Representing driven rotations on the Bloch sphere
In the frame rotating at the drive frequency \(\omega\), any pure state of the two-level subspace corresponds to a unit Bloch vector \(\vec{r} = (\langle X\rangle, \langle Y\rangle, \langle Z\rangle)\) on the Bloch sphere, with \(|0\rangle\) at the north pole (\(z = +1\)) and \(|1\rangle\) at the south pole (\(z = -1\)). A constant drive with Rabi frequency \(\Omega\) and detuning \(\Delta\) generates a rigid precession of \(\vec{r}(t)\) at angular speed \(\Omega_{\mathrm{eff}} = \sqrt{\Omega^2 + \Delta^2}\) around the tilted effective field axis \(\vec{\Omega}_{\mathrm{eff}} = (\Omega, 0, -\Delta)\). When \(\Delta = 0\), the rotation axis lies entirely in the equatorial plane, sweeping the state smoothly through the poles.
Visual animation · Section 13.1
Resonant driving of a two-level transition
Interactive visual model for 13.1 Resonant driving of a two-level transition (Coherent population cycling versus incoherent rate saturation · Separating state preparation, drive calibration, and readout fidelity · Representing driven rotations on the Bloch sphere). Use Animate or Step to follow the physical state evolution and quantitative predictions.
13.2 Rotating-frame Hamiltonian, pulse area, and detuning
13.2.1 Resonant Rabi nutation and the pulse-area angle
Under the rotating-wave approximation (RWA), which drops counter-rotating terms oscillating at \(\omega + \omega_0\) when \(\Omega \ll \omega_0\), the rotating-frame Hamiltonian for a two-level system driven at detuning \(\Delta = \omega - \omega_0\) and Rabi frequency \(\Omega\) is
On exact resonance (\(\Delta = 0\)), time evolution for duration \(t\) depends solely on the dimensionless pulse area \(\Theta = \int_0^t \Omega(t')\,\mathrm{d}t' = \Omega t\). Starting from \(|\psi(0)\rangle = |0\rangle\), the state evolves as
13.2.2 Inversion and superposition pulses on the Bloch sphere
Calibrating the resonant oscillation period \(T_{2\pi} = 2\pi/\Omega\) fixes two fundamental control primitives used across magnetic resonance, atomic clocks, and quantum computing. A \(\pi/2\) pulse (\(\Theta = \Omega t = \pi/2\)) rotates \(|0\rangle\) by \(90^\circ\) onto the equator of the Bloch sphere, preparing the equal-weight superposition \((|0\rangle - i|1\rangle)/\sqrt{2}\) with \(P_1 = 1/2\) and maximum transverse coherence. A \(\pi\) pulse (\(\Theta = \Omega t = \pi\)) rotates \(|0\rangle\) by \(180^\circ\) to \(-i|1\rangle\), inverting the population to \(P_1 = 1\); applied to an arbitrary superposition, a \(\pi\) pulse exchanges the amplitudes of \(|0\rangle\) and \(|1\rangle\), which forms the refocusing element in spin-echo sequences.
13.2.3 Generalized Rabi frequency under nonzero detuning
When the drive is detuned by \(\Delta \neq 0\), the rotation axis tilts out of the equatorial plane by angle \(\vartheta\) with \(\tan\vartheta = \Omega/\Delta\). Exponentiation of \(-iH_{\mathrm{RWA}}t/\hbar\) yields the off-resonant Rabi formula for a system initialized in \(|0\rangle\):
Increasing \(|\Delta|\) produces two linked signatures in a two-dimensional \((t,\Delta)\) chevron plot: the oscillation frequency speeds up from \(\Omega\) to \(\Omega_{\mathrm{eff}} = \sqrt{\Omega^2 + \Delta^2}\), while the maximum transfer amplitude shrinks to the Lorentzian envelope \(\Omega^2/(\Omega^2 + \Delta^2)\).
Visual animation · Section 13.2
Rotating-frame Hamiltonian, pulse area, and detuning
Interactive visual model for 13.2 Rotating-frame Hamiltonian, pulse area, and detuning (Resonant Rabi nutation and the pulse-area angle · Inversion and superposition pulses on the Bloch sphere · Generalized Rabi frequency under nonzero detuning). Use Animate or Step to follow the physical state evolution and quantitative predictions.
13.3 Amplitude scaling, leakage, and control calibration
13.3.1 Verifying linear scaling of Rabi frequency with drive field amplitude
Because the electric-dipole or magnetic-dipole coupling scales with the oscillating field amplitude \(B_1\) or \(E_1\), the resonant Rabi frequency satisfies \(\Omega \propto \sqrt{P_{\mathrm{drive}}}\), where \(P_{\mathrm{drive}}\) is the microwave or laser power delivered to the transition. Plotting the fitted nutation frequency \(\Omega\) against \(\sqrt{P_{\mathrm{drive}}}\) provides a direct linearity audit of the delivery line and transducer: compression at high power diagnoses amplifier saturation or cable heating, while an intercept offset identifies parasitic attenuation.
13.3.2 Breakdown of two-level and rotating-wave approximations
Real physical qubits—such as transmon circuits, hyperfine atomic manifolds, or trapped ions—possess higher-lying levels \(|2\rangle, |3\rangle, \dots\) separated by an anharmonicity \(\alpha = \omega_{12} - \omega_{01}\). When the pulse duration \(t_\pi = \pi/\Omega\) is made too short so that the spectral bandwidth \(\sim \Omega\) becomes comparable to \(|\alpha|\), the drive leaks population out of the \(\{|0\rangle, |1\rangle\}\) computational subspace and induces an AC Stark shift of the resonance frequency. Similarly, if \(\Omega\) approaches a significant fraction of the carrier frequency \(\omega_0\), counter-rotating terms break the rotating-wave approximation and produce Bloch–Siegert shifts and fast ripples on top of the sinusoidal nutation curve.
13.3.3 Why population scans alone do not certify gate fidelity
A resonant Rabi curve measures only the diagonal population \(P_1(t) = \langle 1|\rho(t)|1\rangle = (1 - \langle Z\rangle)/2\) for a single input state \(|0\rangle\). It is insensitive to phase errors around the \(Z\) axis after the pulse, and it cannot distinguish a coherent equatorial state \((|0\rangle - i|1\rangle)/\sqrt{2}\) at \(\Omega t = \pi/2\) from a maximally mixed state \(\frac{1}{2}I\), because both yield \(P_1 = 0.5\). Certifying that a \(\pi/2\) pulse prepares a genuine coherent superposition requires combining Rabi calibration with Ramsey interferometry (Chapter 14) or multi-basis state and process tomography (Chapter 15).
Visual animation · Section 13.3
Amplitude scaling, leakage, and control calibration
Interactive visual model for 13.3 Amplitude scaling, leakage, and control calibration (Verifying linear scaling of Rabi frequency with drive field amplitude · Breakdown of two-level and rotating-wave approximations · Why population scans alone do not certify gate fidelity). Use Animate or Step to follow the physical state evolution and quantitative predictions.
Explore the Apparatus
Adjust the Rabi drive frequency \(\Omega\), the detuning \(\Delta = \omega - \omega_0\), and the pulse duration \(t\) to trace the excited-state population \(P_1(t)\) and the corresponding Bloch-sphere trajectory. Identify the pulse durations that realize \(\pi/2\) superposition pulses and \(\pi\) inversion pulses on resonance, then introduce detuning to observe how the oscillation speeds up while its peak amplitude drops.
Interactive model · Idealized predictions
Rotate a state with a pulse
On a small screen, swipe the diagram to inspect its components. Moving marks illustrate the model; they are not measured trajectories.
Read the values as a table
How does a pulse prepare a chosen quantum state? The text and worked examples remain available without JavaScript. Animation starts only when requested; Step gives a motion-free way to follow the model.
Worked Check
Problem: A two-level transition is driven with resonant Rabi frequency \(\Omega/(2\pi) = 4.0\text{ MHz}\) at a detuning \(\Delta/(2\pi) = 3.0\text{ MHz}\). Find the generalized Rabi frequency \(\Omega_{\mathrm{eff}}/(2\pi)\), the maximum excited-state population \(P_{1,\max}\), and the pulse duration \(t_{\max}\) at which this first peak occurs.
A near-resonant coherent drive couples states \(|0\rangle\) and \(|1\rangle\) with Rabi frequency \(\Omega\) and detuning \(\Delta\), rotating the state vector on the Bloch sphere at generalized rate \(\Omega_{\mathrm{eff}} = \sqrt{\Omega^2 + \Delta^2}\).
On resonance (\(\Delta = 0\)), a \(\pi/2\) pulse prepares an equal superposition and a \(\pi\) pulse inverts population completely, while nonzero detuning caps maximum inversion at \(\Omega^2/(\Omega^2+\Delta^2)\).
Finite relaxation \(T_1\), dephasing \(T_2\), and spatial drive inhomogeneities damp Rabi oscillations toward a mixed steady state at long pulse durations.
Practice Problems
Problem 13.1: Calibrating \(\pi/2\) and \(\pi\) pulse lengths
On exact resonance (\(\Delta = 0\)), the first minimum of \(P_1(t)\) after \(t = 0\) is observed at \(t_{2\pi} = 250\text{ ns}\). What are the resonant Rabi frequency \(\Omega/(2\pi)\), the \(\pi\)-pulse duration \(t_\pi\), and the \(\pi/2\)-pulse duration \(t_{\pi/2}\)?
Hint
The first return to \(P_1 = 0\) occurs at \(\Omega t_{2\pi} = 2\pi\).
Solution
Since \(\Omega t_{2\pi} = 2\pi\), the linear Rabi frequency is \(\Omega/(2\pi) = 1/t_{2\pi} = 1/(250\text{ ns}) = 4.0\text{ MHz}\). The inversion pulse length is \(t_\pi = t_{2\pi}/2 = 125\text{ ns}\), and the superposition pulse length is \(t_{\pi/2} = t_{2\pi}/4 = 62.5\text{ ns}\).
Problem 13.2: Detuning inferred from reduced contrast
A microwave drive with calibrated resonant Rabi rate \(\Omega/(2\pi) = 6.0\text{ MHz}\) is applied to a qubit, and the undamped oscillation peaks at \(P_{1,\max} = 0.36\). Assuming zero SPAM error, what is the magnitude of the detuning \(|\Delta|/(2\pi)\), and what is the observed oscillation period?
Hint
Use \(\Omega^2/\Omega_{\mathrm{eff}}^2 = 0.36\) to find \(\Omega_{\mathrm{eff}}\) first.
Solution
From \(\Omega/\Omega_{\mathrm{eff}} = \sqrt{0.36} = 0.60\), we find \(\Omega_{\mathrm{eff}}/(2\pi) = 6.0 / 0.60 = 10.0\text{ MHz}\). Thus \(|\Delta|/(2\pi) = \sqrt{10.0^2 - 6.0^2} = 8.0\text{ MHz}\), and the oscillation period is \(T = 2\pi/\Omega_{\mathrm{eff}} = 100\text{ ns}\).
Problem 13.3: Distinguishing detuning from readout error
Two experimental traces both show a maximum recorded value \(P_{1,\max} = 0.80\). Trace A is off-resonant by \(\Delta = \Omega/2\) with ideal readout, while Trace B is exactly on resonance (\(\Delta = 0\)) but suffers from a 20% false-zero readout error. How does a scan over both \(t\) and \(\Delta\) distinguish Trace A from Trace B?
Hint
Compare how the oscillation frequency and peak height respond when the drive frequency \(\omega\) is swept across resonance.
Solution
In Trace A, sweeping the drive frequency by \(-\Delta\) slows the oscillation frequency from \(\sqrt{5/4}\,\Omega\) down to \(\Omega\) and restores the peak population to \(1.0\). In Trace B, the oscillation is already at its minimum frequency \(\Omega\), and sweeping \(\omega\) in either direction only increases \(\Omega_{\mathrm{eff}}\) and reduces the peak below \(0.80\), proving that the \(0.20\) ceiling deficit is a readout/preparation offset rather than a detuning error.
Key Takeaways
Coherent driving of a two-level system cycles the excited-state population as \(P_1(t) = (\Omega/\Omega_{\mathrm{eff}})^2\sin^2(\Omega_{\mathrm{eff}}\,t/2)\) with \(\Omega_{\mathrm{eff}} = \sqrt{\Omega^2 + \Delta^2}\).
On resonance (\(\Delta = 0\)), the pulse area \(\Theta = \Omega t\) sets the Bloch rotation angle directly: \(\Theta = \pi/2\) creates an equal coherent superposition and \(\Theta = \pi\) inverts the population.
Nonzero detuning tilts the rotation axis out of the equatorial plane, increasing the oscillation frequency while reducing the maximum population transfer.
Population curves alone calibrate pulse areas but must be paired with phase-sensitive experiments (such as Ramsey fringes or state tomography) to verify equatorial coherence.