Ion Traps

An ion trap confines charged particles with electromagnetic fields while preserving optical access to their internal states. In the most common AMO implementation, a radio-frequency Paul trap supplies dynamic transverse confinement, static electrodes confine the remaining direction, laser cooling reduces the secular motion, and state-dependent fluorescence reveals the internal state.

The compact description

$$\text{electrode voltages} \longrightarrow \text{Mathieu motion} \longrightarrow \text{secular oscillator} \longrightarrow \text{quantized modes}$$

contains several distinct approximations. The exact trajectory in a Paul trap is periodically driven. A time-independent harmonic pseudopotential is a controlled reduction of that dynamics, not the literal electric potential. The fast motion does not vanish when the slow secular motion is cold, and a fluorescing ion is not automatically known to occupy the motional ground state. Keeping those distinctions explicit is essential in precision spectroscopy, quantum information, and motional-state engineering.

Required background. The quantum harmonic oscillator supplies mode quantization, Floquet theory describes periodically driven stability, and laser cooling and atomic selection rules explain optical preparation. Trapped-ion control develops coherent sideband operations and gates.

Why Dynamic Confinement Is Needed

For a charge $Q$ in an electrostatic potential $\Phi(\mathbf r)$, the potential energy is

$$U(\mathbf r)=Q\Phi(\mathbf r).$$

In a charge-free trapping region, Laplace’s equation requires

$$\nabla^2\Phi=0.$$

A positive-definite curvature matrix would have positive trace, contradicting $\nabla^2\Phi=0$. This excludes a harmonic minimum. A flat minimum is not a loophole: the strong maximum principle forbids a nonconstant harmonic potential from having any interior local maximum or minimum. Reversing $Q$ exchanges maxima and minima without evading this obstruction. This is the electrostatic content of Earnshaw’s theorem.

There are two standard ways around the obstruction:

• a Paul trap uses a time-periodic electric quadrupole, so alternating focusing can be dynamically stable;
• a Penning trap combines a static electric quadrupole with a static magnetic field, so the Lorentz force supplies radial confinement.

Paul traps dominate laser-cooled quantum-control experiments because they can provide strong confinement without a large bias magnetic field. Penning traps remain indispensable for precision mass measurements, fundamental-particle studies, clocks, and rotating ion crystals. The remainder of this page focuses on linear Paul traps.

Ideal Quadrupole Paul Trap

Voltage and phase conventions

Consider an ideal two-dimensional quadrupole potential

$$\Phi(x,y,t) = \frac{U_{\mathrm{dc}}-V_{\mathrm{rf}}\cos(\Omega_{\mathrm{rf}}t)} {2r_0^2} \left(x^2-y^2\right).$$

The symbols mean:

Symbol	Meaning
$U_{\mathrm{dc}}$	static quadrupole voltage in this idealized model
$V_{\mathrm{rf}}$	zero-to-peak RF voltage, not peak-to-peak voltage
$\Omega_{\mathrm{rf}}$	RF angular drive frequency
$r_0$	characteristic electrode distance for the chosen potential convention
$Q,m$	ion charge and mass

Changing the sign or phase of the RF term changes intermediate signs but not the stability physics. Real electrodes introduce dimensionless geometry factors and higher multipoles. Therefore a voltage and electrode spacing do not define a trap frequency until the electrostatic model and voltage convention are stated.

Equations of motion

The electric field is $\mathbf E=-\boldsymbol{\nabla}\Phi$. Newton’s equation in the $x$ direction is

$$m\ddot{x} + \frac{Q}{r_0^2} \left[ U_{\mathrm{dc}} - V_{\mathrm{rf}}\cos(\Omega_{\mathrm{rf}}t) \right]x =0.$$

Introduce dimensionless time

$$\tau=\frac{\Omega_{\mathrm{rf}}t}{2}.$$

Then the transverse equations take Mathieu form,

$$\begin{aligned} \frac{d^2x}{d\tau^2} + \left(a_x-2q_x\cos 2\tau\right)x &=0, \\ \frac{d^2y}{d\tau^2} + \left(a_y-2q_y\cos 2\tau\right)y &=0, \end{aligned}$$

with

$$\begin{aligned} a_x &= \frac{4QU_{\mathrm{dc}}} {mr_0^2\Omega_{\mathrm{rf}}^2}, & q_x &= \frac{2QV_{\mathrm{rf}}} {mr_0^2\Omega_{\mathrm{rf}}^2}, \\ a_y&=-a_x, & q_y&=-q_x. \end{aligned}$$

The opposite signs express alternating focusing: when the instantaneous field focuses along $x$, it defocuses along $y$, and half an RF period later the roles reverse.

Stability is a Floquet property

Inside a stable region, a real Mathieu solution can be written

$$x(\tau)=c\,e^{i\beta_x\tau}P_x(\tau) +c^*e^{-i\beta_x\tau}P_x^*(\tau), \qquad P_x(\tau+\pi)=P_x(\tau),$$

with real characteristic exponent $\beta_x$. One can test stability directly by propagating two independent initial conditions through one RF period. The resulting two-by-two monodromy matrix $M$ maps $(x,x')$ forward by $\pi$ in $\tau$; its determinant is one. For real coefficients,

$$|\operatorname{tr}M|<2$$

gives two distinct unit-modulus multipliers and bounded motion. At $|\operatorname{tr}M|=2$, a real exponent alone is insufficient: a defective matrix produces a second solution growing linearly with time. Thus stability depends on the full periodic equation, including its boundary behavior.

Along $a=0$, the first open stability interval is approximately

$$0<|q|<0.908.$$

At $a=q=0$ the particle is free, not confined. The exact characteristic boundaries are described in the NIST Mathieu stability chart.

This boundary is not a recommended operating point. Quantum-control experiments commonly use smaller $|q|$, often of order $0.1$ to $0.4$, to reduce micromotion and make the secular approximation accurate. Exact stability must be evaluated with the actual DC curvatures and electrode geometry when $a$ or $q$ is not small.

Top: direct integration of $x''-2q\cos(2\tau)x=0$ at $q=0.3$, $x(0)=0.85$, $x'(0)=0$, compared with $\cos(q\tau/\sqrt2)[1-(q/2)\cos2\tau]$ and its slow factor. Bottom: $\beta=\pi^{-1}\arccos(\operatorname{tr}M/2)$ from one-period propagation, compared with $q/\sqrt2$. Micromotion and the error in secular frequency are different effects; the latter grows toward the stability edge.

Secular Motion and the Pseudopotential

Small-parameter solution

For $|a_x|\ll1$ and $q_x^2\ll1$, away from the stability boundary, the leading characteristic exponent is

$$\beta_x \simeq \sqrt{a_x+\frac{q_x^2}{2}},$$

so the slow secular angular frequency is

$$\omega_x \simeq \frac{\Omega_{\mathrm{rf}}}{2} \sqrt{a_x+\frac{q_x^2}{2}}.$$

To first order in $q_x$, a representative trajectory is

$$x(t) \simeq X\cos(\omega_xt+\varphi) \left[ 1-\frac{q_x}{2} \cos(\Omega_{\mathrm{rf}}t) \right].$$

This expression displays two time scales:

• $X\cos(\omega_xt+\varphi)$ is the slow secular motion;
• the factor oscillating at $\Omega_{\mathrm{rf}}$ is micromotion.

The sign of the modulation depends on the RF phase convention. Its magnitude relative to the secular displacement is $|q_x|/2$ at this order.

Effective potential energy

Let $\mathbf E_{\mathrm{rf}}(\mathbf r)$ be the amplitude of the RF electric field, so that the rapidly oscillating field is $\mathbf E_{\mathrm{rf}}(\mathbf r)\cos\Omega_{\mathrm{rf}}t$. Averaging over the fast motion gives the ponderomotive or pseudopotential energy

$$U_{\mathrm{ps}}(\mathbf r) = \frac{Q^2} {4m\Omega_{\mathrm{rf}}^2} \left| \mathbf E_{\mathrm{rf}}(\mathbf r) \right|^2.$$

For the ideal quadrupole,

$$U_{\mathrm{ps}}(x,y) = \frac{Q^2V_{\mathrm{rf}}^2} {4m\Omega_{\mathrm{rf}}^2r_0^4} \left(x^2+y^2\right).$$

Matching this to

$$U_{\mathrm{ps}} = \frac{m\omega_{\mathrm{rf}}^2}{2} \left(x^2+y^2\right)$$

gives

$$\omega_{\mathrm{rf}} = \frac{|Q|V_{\mathrm{rf}}} {\sqrt{2}\,m\Omega_{\mathrm{rf}}r_0^2} = \frac{|q|\Omega_{\mathrm{rf}}} {2\sqrt{2}}$$

when $a=0$.

The effective confinement becomes weaker when the drive frequency is increased at fixed voltage, because the ion has less time to acquire the correlated fast displacement that produces the average restoring force. Holding $q$ fixed while increasing $\Omega_{\mathrm{rf}}$, however, requires $V_{\mathrm{rf}}\propto\Omega_{\mathrm{rf}}^2$ and increases the secular frequency. Statements about scaling must say which control variable is held fixed.

Domain of validity

The pseudopotential is reliable when:

1. $|q|$ and $|a|$ are sufficiently small;
2. secular amplitudes remain in the nearly quadrupolar region;
3. the secular spectrum is well separated from the RF drive and dangerous nonlinear resonances;
4. applied forces vary slowly compared with $\Omega_{\mathrm{rf}}$;
5. micromotion-sensitive observables are either averaged correctly or modeled with the full periodic dynamics.

It can fail near stability boundaries, in strongly anharmonic electrode fields, during rapid transport, for large Coulomb crystals, or when laser coupling resolves RF sidebands. The effective potential predicts secular confinement; it does not erase the periodic kinetic energy.

Micromotion

Intrinsic micromotion

Intrinsic micromotion is the RF modulation tied to secular displacement. In the lowest-order solution,

$$x_{\mathrm{imm}}(t) \simeq - \frac{q_x}{2} x_{\mathrm{sec}}(t) \cos(\Omega_{\mathrm{rf}}t).$$

Cooling the secular oscillator toward its ground state reduces the spatial extent sampled by this term, but the exact trapped state remains a Floquet state with periodic motion. It is therefore imprecise to say that ground-state cooling removes all micromotion.

Excess micromotion from a displaced equilibrium

A static stray field $E_s$ displaces an ion from the RF null. In one harmonic direction,

$$x_d = \frac{QE_s} {m\omega_x^2}.$$

The corresponding first-order excess-micromotion amplitude is

$$x_{\mathrm{emm}} \simeq \frac{|q_x|}{2}|x_d|.$$

This motion persists even if the secular state has $\bar n\ll1$. It can produce:

• first-order Doppler modulation and RF sidebands;
• second-order Doppler shifts;
• RF AC Stark shifts;
• altered laser-cooling forces;
• collision energy that cannot be represented by a secular temperature;
• gate and spectroscopy errors.

Compensation electrodes apply static fields that move the equilibrium back toward the RF null. A complete three-dimensional compensation also addresses RF phase imbalance between electrodes, which can create driven motion along a nominally field-free axis. Static compensation alone generally cannot cancel that component; the RF feed phases or electrode amplitudes must also be corrected.

Detecting and compensating micromotion

Three established diagnostics probe different projections and systematics:

Diagnostic	Observable	Important caveat
position versus confinement strength	displacement changes as the secular frequency is varied	requires calibrated imaging and distinguishes static-force directions
resolved RF sidebands	modulation index on a narrow optical transition	optical phase and AC Stark modulation can contribute
photon correlation	fluorescence modulation relative to the RF phase	depends on cooling-transition detuning, saturation, and laser direction

A single beam has no first-order Doppler sensitivity to motion perpendicular to its wavevector; other signals, such as RF Stark modulation, need not share that geometric null. Micromotion should be bounded along enough independent directions, under the same voltages used for the experiment, and rechecked after loading or charging events. Berkeland et al. give the classic quantitative treatment of these methods.

Linear Paul Traps

A linear Paul trap uses an approximately two-dimensional RF quadrupole for radial confinement and static end electrodes for axial confinement. Near the center, write the endcap potential as

$$\Phi_{\mathrm{end}} \simeq \frac{\kappa U_{\mathrm{end}}}{z_0^2} \left[ z^2-\frac{x^2+y^2}{2} \right],$$

where $\kappa$ is a dimensionless geometry factor. The axial frequency is

$$\omega_z^2 = \frac{2Q\kappa U_{\mathrm{end}}} {mz_0^2}.$$

Laplace’s equation forces the accompanying radial curvature to be defocusing. In an ideal symmetric trap,

$$\omega_x^2 \simeq \omega_{\mathrm{rf}}^2 - \frac{\omega_z^2}{2}, \qquad \omega_y^2 \simeq \omega_{\mathrm{rf}}^2 - \frac{\omega_z^2}{2}.$$

Additional DC quadrupoles intentionally split $\omega_x$ and $\omega_y$ or rotate the radial principal axes. The sum of the three static electric curvatures must still vanish. Thus axial confinement is not an independent addition to the radial problem: it spends part of the radial stability margin.

Practical trap geometries include:

• three-dimensional ring-and-endcap traps;
• four-rod linear traps;
• segmented linear traps for axial shaping and ion transport;
• surface-electrode traps fabricated in one plane;
• cryogenic traps used to reduce some noise and vacuum limitations.

The ideal Mathieu model remains the organizing approximation, but numerical electrostatics is normally required to convert electrode voltages into curvatures, RF fields, anharmonicities, and transport waveforms.

Quantized Secular Motion

One ion

Near a stable equilibrium, the secular Hamiltonian is

$$H_{\mathrm{mot}} = \sum_{\alpha=x,y,z} \hbar\omega_\alpha \left( a_\alpha^\dagger a_\alpha+\frac{1}{2} \right).$$

For one mode,

$$x = x_0(a+a^\dagger), \qquad x_0 = \sqrt{\frac{\hbar}{2m\omega}}.$$

The oscillator quantum number $n$ refers to the secular mode in the effective description. The exact periodically driven quantum state has RF-periodic micromotion dressing. This distinction matters when comparing oscillator tomography, Doppler shifts, and full time-resolved trajectories.

A laser with wavevector projection $k_{\mathrm{eff}}$ couples to motion through

$$e^{ik_{\mathrm{eff}}x} = \exp\!\left[ i\eta(a+a^\dagger) \right], \qquad \eta=k_{\mathrm{eff}}x_0.$$

The Lamb–Dicke condition

$$\eta\sqrt{2\bar n+1}\ll1$$

means that the optical phase changes little across the occupied wavepacket. It is a condition on mode, beam geometry, and motional state, not a universal property of an ion species.

Trap depth is not an oscillator frequency

The local curvature determines $\omega$, whereas escape is controlled by the global time-dependent electrode potential, nonlinear resonances, collisions, and finite electrode apertures. A harmonic fit alone cannot establish trap depth. Conversely, a large nominal pseudopotential depth does not ensure low heating or good optical control.

Coulomb Crystals and Motional Modes

For $N$ equal ions in a harmonic secular potential, a useful effective potential energy is

$$U = \sum_{i=1}^{N} \frac{m}{2} \left( \omega_x^2x_i^2 + \omega_y^2y_i^2 + \omega_z^2z_i^2 \right) + \sum_{i<j} \frac{Q^2} {4\pi\epsilon_0 |\mathbf r_i-\mathbf r_j|}.$$

At sufficiently low secular temperature, the ions localize near equilibrium positions that minimize $U$. The term Coulomb crystal describes this ordered configuration; it does not imply electronic crystalline bonding.

Linear chains

When radial confinement is much stronger than axial confinement, the equilibrium positions form a chain along the RF null. If radial confinement is reduced, a transverse normal-mode frequency softens and the chain can undergo a zigzag structural transition. Large crystals can sample anharmonicity and RF fields away from the null, so the single-particle pseudopotential test must be repeated for the many-ion configuration.

Normal-mode construction

Let $u_{i\alpha}$ be a small displacement from equilibrium. Expand to second order:

$$U \simeq U_0 + \frac{1}{2} \sum_{i\alpha,j\beta} K_{i\alpha,j\beta} u_{i\alpha}u_{j\beta},$$

where

$$K_{i\alpha,j\beta} = \left. \frac{\partial^2U} {\partial u_{i\alpha}\partial u_{j\beta}} \right|_{\mathrm{eq}}.$$

For equal masses, diagonalizing $K/m$ gives eigenvalues $\omega_m^2$ and orthonormal eigenvectors $b_{i\alpha,m}$. Quantization gives

$$u_{i\alpha} = \sum_m b_{i\alpha,m} \sqrt{\frac{\hbar}{2m\omega_m}} \left(a_m+a_m^\dagger\right).$$

Each laser couples to a mode with a participation factor $\mathbf k_{\mathrm{eff}}\cdot\mathbf b_{i,m}$. A mode can be cold yet nearly invisible to a poorly oriented probe, or strongly driven by spatially correlated electric-field noise. Quoting one “ion temperature” can hide this mode dependence.

Two-ion benchmark

For two equal ions at $z=\pm z_e$, force balance gives

$$z_e^3 = \frac{Q^2} {16\pi\epsilon_0m\omega_z^2}.$$

The axial center-of-mass and stretch frequencies are

$$\omega_{\mathrm{COM}}=\omega_z, \qquad \omega_{\mathrm{str}}=\sqrt{3}\,\omega_z.$$

These values are a useful calibration benchmark. Significant disagreement can signal unequal masses, anharmonic confinement, incorrect frequency assignment, or a geometry outside the assumed one-dimensional equilibrium.

Internal States

Confinement acts on the ion’s center of mass, while spectroscopy and quantum control usually use electronic, fine-structure, hyperfine, or Zeeman states. The mapping from an atomic spectrum to an effective two-level system is an engineering choice.

Encoding	Typical control	Strengths	Leading qualifications
optical qubit	narrow electric-quadrupole or octupole transition	direct optical addressing; useful for clocks and coherent control	laser coherence, finite excited-state lifetime, off-resonant levels
hyperfine clock qubit	microwaves or stimulated Raman fields	long coherence near field-insensitive points	Raman scattering, differential light shifts, extra repumping levels
Zeeman qubit	RF, microwave, or Raman fields	simple level structure and control	first-order magnetic-field sensitivity unless protected
metastable shelving state	optical excitation and fluorescence mapping	large bright–dark contrast	spontaneous decay and imperfect state transfer

Common species include $^9\mathrm{Be}^+$, $^{25}\mathrm{Mg}^+$, $^{40}\mathrm{Ca}^+$, $^{88}\mathrm{Sr}^+$, $^{138}\mathrm{Ba}^+$, and ytterbium isotopes. Species choice determines wavelengths, level complexity, branching ratios, isotope structure, clock transitions, and available photoionization paths. It does not by itself determine coherence or measurement fidelity; the magnetic environment, lasers, collection optics, and control protocol remain part of the system.

Hyperfine Structure, The Zeeman Effect in Atoms, and Atomic Selection Rules provide the atomic-structure background.

Loading and Initial Preparation

A typical preparation sequence is:

1. produce neutral atoms from an oven, ablation source, or atomic beam;
2. ionize the desired isotope, preferably with isotope-selective photoionization;
3. capture the ion within the trap’s stable phase-space region;
4. Doppler cool and detect fluorescence;
5. compensate static fields and characterize micromotion;
6. optically pump the internal state;
7. cool selected secular modes further if the protocol requires it.

Electron-impact ionization can load many species but often adds charging, background gas, and poor isotope selectivity. Resonant photoionization offers more control but can still charge exposed dielectrics. Loading therefore changes the trap environment and may invalidate a previously measured micromotion compensation.

Background-gas collisions can reorder a mixed-species chain, transfer motional energy, cause chemical reactions, or eject an ion. A long storage lifetime is evidence about rare loss processes; it is not a direct measurement of the electric-field noise near a secular frequency that controls motional heating.

Cooling the Secular Modes

Doppler cooling

Radiation pressure from red-detuned light provides velocity-dependent damping while absorption and spontaneous emission supply momentum diffusion. For the usual symmetric recoil geometry in the low-saturation two-level model, the familiar minimum temperature is

$$k_BT_D = \frac{\hbar\Gamma}{2},$$

where $\Gamma$ is the excited-state population decay rate in angular-frequency units. Real ions require repump lasers, polarization control, magnetic fields, and treatment of multilevel dark states. The ideal formula is therefore a scale, not a universal thermometer.

For a thermal oscillator,

$$\bar n = \frac{1} {\exp(\hbar\omega/k_BT)-1}.$$

When $k_BT\gg\hbar\omega$,

$$\bar n \simeq \frac{k_BT}{\hbar\omega}.$$

Because typical optical linewidths can greatly exceed secular frequencies, Doppler cooling often leaves $\bar n\gg1$ even though the ion is localized on a microscopic scale.

Resolved-sideband cooling

If motional sidebands are spectrally resolved,

$$\omega_m\gg\Gamma_{\mathrm{eff}},$$

one can drive a red sideband

$$|g,n\rangle \longrightarrow |e,n-1\rangle$$

and use dissipative repumping to return the internal state while usually preserving the reduced motional number in the Lamb–Dicke regime. Repetition accumulates population near $n=0$, where the red sideband vanishes.

Ground-state cooling was demonstrated in one dimension by Diedrich et al. and in all three dimensions with Raman sideband cooling by Monroe et al. The excitation linewidth and drive strength must suppress off-resonant carrier and blue-sideband excitation. A separate repumping pulse may be broad; its absorption and emission recoils have their own Lamb–Dicke parameters. See sub-Doppler cooling for the rates and cooling limits.

Other cooling strategies

• EIT cooling can cool multiple modes over a broader engineered spectral window than a single narrow red sideband.
• Sympathetic cooling uses a second ion species to remove motion without directly scattering photons from a protected spectroscopy or logic ion.
• Continuous cooling can stabilize spectator modes but may introduce scattering and differential forces during coherent operations.

Cooling is mode specific. A reported $\bar n$ should identify the mode, diagnostic, fit model, and delay between cooling and use.

Measuring Internal and Motional States

State-dependent fluorescence

The standard internal-state measurement maps one state to a nearly closed cycling transition and leaves another state dark or shelved. During a detection interval, the apparatus records a photon count $n$ or a time-resolved sequence of counts.

For weak photon collection and counting intervals long compared with optical correlation times, a useful static approximation is

$$\begin{aligned} P(n|b) &= e^{-\mu_b} \frac{\mu_b^n}{n!}, \\ P(n|d) &= e^{-\mu_d} \frac{\mu_d^n}{n!}, \end{aligned}$$

with $\mu_b\gg\mu_d$. Single-ion emission is antibunched, so these Poisson laws are an approximation even without state changes. A threshold or likelihood-ratio rule classifies the state. This model must be enlarged when the dark state decays, off-resonant pumping changes the state during detection, the collection rate drifts, or neighboring ions overlap on a camera.

Electron shelving made individual quantum jumps directly visible as bright and dark fluorescence intervals. Modern time-resolved likelihood methods can outperform a fixed count threshold because the timing of photons helps distinguish a genuinely bright state from a dark state that decays partway through detection.

The measurement backaction and trajectory interpretation are developed in Trapped Ions and Quantum-Jump Trajectories.

Motional diagnostics

Common diagnostics include:

• secular-frequency spectroscopy with a weak electric or optical drive;
• red-to-blue sideband asymmetry near the ground state;
• Rabi oscillations whose $n$-dependent frequencies reveal a number distribution;
• controlled displacement followed by internal-state mapping;
• imaging of equilibrium positions and thermally broadened spatial distributions;
• heating-rate measurements after a variable delay without cooling.

For any diagonal motional number distribution with finite mean, the Lamb–Dicke weak-excitation matrix elements give $I_{\mathrm{red}}\propto \sum_n nP_n$ and $I_{\mathrm{blue}}\propto\sum_n(n+1)P_n$. Hence

$$\frac{I_{\mathrm{red}}} {I_{\mathrm{blue}}} = \frac{\bar n} {\bar n+1}.$$

Thermality is unnecessary for this mean-occupation relation. It is needed to infer a temperature or $P_0=1/(1+\bar n)$ from the mean alone. Probe saturation, unequal red/blue response, off-resonant carrier excitation, micromotion sidebands, mode overlap, and state-preparation errors can bias the inferred $\bar n$.

Heating and Environmental Coupling

For a single ion’s mode polarized along $\mathbf e_m$, weak classical stationary electric-field noise couples through $-Qx\,\delta E_m(t)$. Specify the one-sided angular-frequency spectrum for $\omega>0$ as

$$S_{E,m}(\omega)=2\int_{-\infty}^{\infty}d\tau\, e^{i\omega\tau}\langle\delta E_m(t+\tau)\delta E_m(t)\rangle.$$

It obeys $\langle\delta E_m^2\rangle=(2\pi)^{-1}\int_0^\infty S_{E,m}(\omega)d\omega$. In the weak-noise oscillator approximation,

$$\dot{\bar n}_m = \frac{Q^2} {4m\hbar\omega_m} S_{E,m}(\omega_m).$$

The numerical prefactor changes with one-sided versus two-sided and angular- frequency versus ordinary-frequency conventions. A heating-rate result should therefore report the convention before converting to $S_E$. A quantum bath requires separate unsymmetrized spectra for upward and downward transitions; the classical heating formula does not describe relaxation to an arbitrary cold bath. Collective modes also sample spatial cross-correlations of the fields at different ions.

Technical pickup, electrode-voltage noise, dielectric charging, fluctuating surface potentials, RF noise near secular sidebands, and collisions can all contribute. The extensive evidence and open questions concerning surface-related electric-field noise are discussed in the open-system trapped-ion page and the review by Brownnutt et al.

Worked Example: One Calcium Ion

Consider an idealized $^{40}\mathrm{Ca}^+$ ion with

$$\begin{gathered} Q=e, \qquad m=39.9626\,u, \\ r_0=0.500\ \mathrm{mm}, \qquad V_{\mathrm{rf}}=200\ \mathrm{V}, \\ \frac{\Omega_{\mathrm{rf}}}{2\pi} = 20.0\ \mathrm{MHz}, \qquad U_{\mathrm{dc}}=0. \end{gathered}$$

Here $V_{\mathrm{rf}}$ is the zero-to-peak voltage in the ideal quadrupole potential. The Mathieu parameter is

$$q = \frac{2eV_{\mathrm{rf}}} {mr_0^2\Omega_{\mathrm{rf}}^2} = 0.2446.$$

The leading pseudopotential estimate of the RF-only secular frequency is

$$\frac{\omega_{\mathrm{rf}}}{2\pi} = \frac{q\Omega_{\mathrm{rf}}} {4\sqrt{2}\,\pi} = 1.730\ \mathrm{MHz}.$$

Direct one-period Mathieu propagation at $q=0.2446$ gives $\beta\simeq0.17505$, or $1.7505\ \mathrm{MHz}$: about $1.2\%$ above the leading estimate. The following numbers consistently use the pseudopotential approximation.

The associated zero-point length is

$$x_0 = \sqrt{\frac{\hbar}{2m\omega_{\mathrm{rf}}}} = 8.55\ \mathrm{nm}.$$

If static end electrodes produce $\omega_z/(2\pi)=0.500\ \mathrm{MHz}$, the ideal radial defocusing estimate gives

$$\frac{\omega_r}{2\pi} \simeq \frac{1}{2\pi} \sqrt{ \omega_{\mathrm{rf}}^2-\frac{\omega_z^2}{2} } = 1.693\ \mathrm{MHz}.$$

Now suppose a transverse static stray field is $E_s=10.0\ \mathrm{V\,m^{-1}}$. Using the RF-only curvature for a transparent scale estimate,

$$x_d = \frac{eE_s} {m\omega_{\mathrm{rf}}^2} = 204\ \mathrm{nm},$$

and

$$x_{\mathrm{emm}} \simeq \frac{q}{2}x_d = 25.0\ \mathrm{nm}.$$

The excess-micromotion amplitude is about three times the secular zero-point length despite the apparently modest static field. This comparison explains why micromotion compensation is a precision calibration, not merely a visual centering step.

These numbers are not predictions for a particular electrode structure. Replacing the ideal quadrupole by a real trap requires geometry factors from electrostatic modeling, measured secular frequencies, RF-voltage calibration, and a check for phase-driven micromotion.

Exercises

Exercise 1: Derive the Mathieu parameters

Starting from

$$\Phi(x,y,t) = \frac{U_{\mathrm{dc}}-V_{\mathrm{rf}}\cos\Omega_{\mathrm{rf}}t} {2r_0^2} (x^2-y^2),$$

derive the Mathieu equations for $x$ and $y$. Explain why the signs of both $a$ and $q$ reverse between the two directions.

Solution

The fields are

$$E_x = - \frac{U_{\mathrm{dc}}-V_{\mathrm{rf}}\cos\Omega_{\mathrm{rf}}t} {r_0^2}x,$$

and

$$E_y = + \frac{U_{\mathrm{dc}}-V_{\mathrm{rf}}\cos\Omega_{\mathrm{rf}}t} {r_0^2}y.$$

Using $m\ddot{x}=QE_x$ gives

$$m\ddot{x} + \frac{Q}{r_0^2} \left( U_{\mathrm{dc}} - V_{\mathrm{rf}}\cos\Omega_{\mathrm{rf}}t \right)x =0.$$

Set $\tau=\Omega_{\mathrm{rf}}t/2$, so $d^2/dt^2=(\Omega_{\mathrm{rf}}^2/4)d^2/d\tau^2$. Then

$$\frac{d^2x}{d\tau^2} + \left[ \frac{4QU_{\mathrm{dc}}} {mr_0^2\Omega_{\mathrm{rf}}^2} - \frac{4QV_{\mathrm{rf}}} {mr_0^2\Omega_{\mathrm{rf}}^2} \cos2\tau \right]x =0.$$

Comparison with $x''+(a_x-2q_x\cos2\tau)x=0$ gives

$$a_x = \frac{4QU_{\mathrm{dc}}} {mr_0^2\Omega_{\mathrm{rf}}^2}, \qquad q_x = \frac{2QV_{\mathrm{rf}}} {mr_0^2\Omega_{\mathrm{rf}}^2}.$$

Because $\Phi$ contains $-y^2$ where it contains $+x^2$, the force curvature changes sign:

$$a_y=-a_x, \qquad q_y=-q_x.$$

This is the mathematical statement that the instantaneous quadrupole focuses one transverse direction while defocusing the other.

Exercise 2: Recover the pseudopotential frequency

For the ideal quadrupole with $U_{\mathrm{dc}}=0$, show that the pseudopotential gives

$$\omega_{\mathrm{rf}} = \frac{|q|\Omega_{\mathrm{rf}}}{2\sqrt{2}}.$$

Why does $\omega_{\mathrm{rf}}$ decrease with increasing $\Omega_{\mathrm{rf}}$ if $V_{\mathrm{rf}}$ is held fixed?

Solution

The RF-field amplitude is

$$\mathbf E_{\mathrm{rf}} = \frac{V_{\mathrm{rf}}}{r_0^2} (x\,\mathbf e_x-y\,\mathbf e_y),$$

so

$$|\mathbf E_{\mathrm{rf}}|^2 = \frac{V_{\mathrm{rf}}^2}{r_0^4} (x^2+y^2).$$

The pseudopotential energy is

$$U_{\mathrm{ps}} = \frac{Q^2V_{\mathrm{rf}}^2} {4m\Omega_{\mathrm{rf}}^2r_0^4} (x^2+y^2).$$

Equating its coefficient to $m\omega_{\mathrm{rf}}^2/2$ gives

$$\omega_{\mathrm{rf}} = \frac{|Q|V_{\mathrm{rf}}} {\sqrt{2}\,m\Omega_{\mathrm{rf}}r_0^2}.$$

Since

$$|q| = \frac{2|Q|V_{\mathrm{rf}}} {mr_0^2\Omega_{\mathrm{rf}}^2},$$

substitution yields

$$\omega_{\mathrm{rf}} = \frac{|q|\Omega_{\mathrm{rf}}}{2\sqrt{2}}.$$

At fixed $V_{\mathrm{rf}}$, the fast displacement induced during one RF cycle decreases as the drive becomes faster, so the averaged restoring energy scales as $\Omega_{\mathrm{rf}}^{-2}$ and the frequency as $\Omega_{\mathrm{rf}}^{-1}$. At fixed $q$, by contrast, $V_{\mathrm{rf}}\propto\Omega_{\mathrm{rf}}^2$ and $\omega_{\mathrm{rf}}\propto\Omega_{\mathrm{rf}}$.

Exercise 3: When a two-ion chain bends

Two equal ions have axial equilibrium positions $\pm z_e$ and radial single-ion secular frequency $\omega_r$. Derive the radial center-of-mass and rocking frequencies. At what anisotropy does the linear chain lose stability in this two-ion pseudopotential model?

Solution

Let $C=Q^2/(4\pi\epsilon_0)$. For small radial displacements,

$$\frac{C}{\sqrt{(2z_e)^2+(x_1-x_2)^2}} =\frac{C}{2z_e}-\frac{C(x_1-x_2)^2}{16z_e^3}+O(x^4).$$

Using $C/(4mz_e^3)=\omega_z^2$ gives the radial Hessian

$$\frac{K_r}{m}= \begin{pmatrix} \omega_r^2-\omega_z^2/2&\omega_z^2/2\\ \omega_z^2/2&\omega_r^2-\omega_z^2/2 \end{pmatrix}.$$

Its symmetric and antisymmetric eigenvectors have frequencies $\omega_{\mathrm{COM},r}=\omega_r$ and $\omega_{\mathrm{rock}}=\sqrt{\omega_r^2-\omega_z^2}$. Coulomb repulsion softens the relative transverse motion, while a common displacement leaves the separation unchanged. At $\omega_r=\omega_z$ the chain orientation is marginal; below it the axial chain is unstable. Longer chains have different critical anisotropies, determined by their full Hessians.

Exercise 4: Excess micromotion from a stray field

An ion with secular frequency $\omega_x$ experiences a static field $E_s$. Derive its displaced equilibrium and first-order excess-micromotion amplitude. If $\omega_x$ is doubled while $q$ and $E_s$ are held fixed, by what factor does the excess-micromotion amplitude change?

Solution

The slow effective force is

$$F_{\mathrm{eff}} = - m\omega_x^2x+QE_s.$$

Setting it to zero gives

$$x_d = \frac{QE_s}{m\omega_x^2}.$$

The small-$q$ Mathieu solution modulates a static displacement by a fractional amplitude $|q|/2$, so

$$x_{\mathrm{emm}} \simeq \frac{|q|}{2} \frac{|Q E_s|}{m\omega_x^2}.$$

At fixed $q$ and $E_s$, doubling $\omega_x$ reduces both $x_d$ and $x_{\mathrm{emm}}$ by a factor of four. In a real trap, changing $\omega_x$ may also change $q$, the principal axes, or the stray field, so the experimental scaling test must track those quantities.

Exercise 5: Two-ion equilibrium and normal modes

Two equal ions are confined to the $z$ axis by a harmonic potential of frequency $\omega_z$. Derive their equilibrium positions and the axial center-of-mass and stretch frequencies.

Solution

Let the ions sit at $z_1=z_e$ and $z_2=-z_e$. With $C=Q^2/(4\pi\epsilon_0)$, the potential is

$$U(z_e) = m\omega_z^2z_e^2 + \frac{C}{2z_e}.$$

Force balance gives

$$\frac{dU}{dz_e} = 2m\omega_z^2z_e - \frac{C}{2z_e^2} =0,$$

and hence

$$z_e^3 = \frac{C}{4m\omega_z^2} = \frac{Q^2} {16\pi\epsilon_0m\omega_z^2}.$$

For small displacements $u_1,u_2$, the axial Hessian is

$$K = m\omega_z^2 \begin{pmatrix} 2 & -1\\ -1 & 2 \end{pmatrix}.$$

The normalized eigenvectors are proportional to $(1,1)$ and $(1,-1)$, with eigenvalues $m\omega_z^2$ and $3m\omega_z^2$. Therefore

$$\omega_{\mathrm{COM}}=\omega_z, \qquad \omega_{\mathrm{str}}=\sqrt{3}\,\omega_z.$$

The Coulomb force does not change the center-of-mass frequency because a uniform translation leaves the separation unchanged.

Exercise 6: Why Doppler cooling need not reach the ground state

For the symmetric-recoil two-level Doppler benchmark with $\Gamma/(2\pi)=21.6\ \mathrm{MHz}$, calculate the ideal Doppler temperature. Estimate the thermal occupations of secular modes at $1.70\ \mathrm{MHz}$ and $0.500\ \mathrm{MHz}$. Use the exact Bose–Einstein oscillator expression.

Solution

The ideal Doppler temperature is

$$T_D = \frac{\hbar\Gamma}{2k_B} = \frac{h(21.6\ \mathrm{MHz})}{2k_B} \simeq 5.18\times10^{-4}\ \mathrm{K}.$$

For a mode of ordinary frequency $f$,

$$\bar n = \frac{1} {\exp(hf/k_BT_D)-1}.$$

At $f=1.70\ \mathrm{MHz}$,

$$\frac{hf}{k_BT_D} \simeq 0.157, \qquad \bar n\simeq5.90.$$

At $f=0.500\ \mathrm{MHz}$,

$$\frac{hf}{k_BT_D} \simeq 0.0463, \qquad \bar n\simeq21.1.$$

Thus sub-millikelvin cooling can still leave many secular quanta. The result uses an ideal two-level Doppler limit; multilevel structure, polarization, micromotion, recoil, and laser parameters change the actual steady state.

Exercise 7: Sideband-asymmetry thermometry

A weak probe measures an integrated red-to-blue first-sideband ratio $R=0.120$. Assuming a thermal mode and the ideal Lamb–Dicke weak-probe relation, infer $\bar n$ and the ground-state population $P_0$. Name three effects that could invalidate the inference.

Solution

For any diagonal mode distribution in this weak-probe limit,

$$R = \frac{\bar n}{\bar n+1}.$$

Solving gives

$$\bar n = \frac{R}{1-R} = \frac{0.120}{0.880} = 0.136.$$

The thermal number distribution is

$$P_n = \frac{\bar n^n}{(\bar n+1)^{n+1}},$$

so

$$P_0 = \frac{1}{1+\bar n} = 0.880.$$

Possible failures include probe saturation, off-resonant carrier excitation, overlapping modes, micromotion sidebands, state-preparation error, unequal red and blue detection transfer functions, and heating during the scan. A nonthermal distribution alone does not change the inferred mean, but invalidates this inference of $P_0$. Sideband asymmetry is powerful because many coupling factors cancel, but only within the model that produces the ratio.

Exercise 8: Micromotion and a clock shift

Suppose excess micromotion is $x(t)=A\cos\Omega_{\mathrm{rf}}t$ with $A=25.0\ \mathrm{nm}$ and $\Omega_{\mathrm{rf}}/(2\pi)=20.0\ \mathrm{MHz}$. Find its cycle-averaged kinetic energy and special-relativistic clock shift. Does secular ground-state cooling remove this contribution?

Solution

The mean squared speed is $\langle v^2\rangle=A^2\Omega_{\mathrm{rf}}^2/2$. Therefore

$$\langle K_{\mathrm{emm}}\rangle=\frac{mA^2\Omega_{\mathrm{rf}}^2}{4}, \qquad \frac{\delta\nu}{\nu}=-\frac{\langle v^2\rangle}{2c^2} =-\frac{A^2\Omega_{\mathrm{rf}}^2}{4c^2} \simeq-2.75\times10^{-17}.$$

For the calcium mass used above, $\langle K_{\mathrm{emm}}\rangle/k_B \simeq11.9\ \mathrm{mK}$. This energy scale is not a secular temperature: the motion is coherently driven at the RF frequency. Ground-state cooling does not remove a displacement from the RF null. Clock measurements must also account for the RF Stark shift, which is distinct from this kinetic contribution; see optical clocks.

References

• J. C. Bergquist, R. G. Hulet, W. M. Itano, and D. J. Wineland, “Observation of quantum jumps in a single atom,” Physical Review Letters 57, 1699–1702 (1986), doi:10.1103/PhysRevLett.57.1699.
• D. J. Berkeland, J. D. Miller, J. C. Bergquist, W. M. Itano, and D. J. Wineland, “Minimization of ion micromotion in a Paul trap,” Journal of Applied Physics 83, 5025–5033 (1998), doi:10.1063/1.367318.
• L. S. Brown and G. Gabrielse, “Geonium theory: Physics of a single electron or ion in a Penning trap,” Reviews of Modern Physics 58, 233–311 (1986), doi:10.1103/RevModPhys.58.233.
• M. Brownnutt, M. Kumph, P. Rabl, and R. Blatt, “Ion-trap measurements of electric-field noise near surfaces,” Reviews of Modern Physics 87, 1419–1482 (2015), doi:10.1103/RevModPhys.87.1419.
• C. D. Bruzewicz, J. Chiaverini, R. McConnell, and J. M. Sage, “Trapped-ion quantum computing: Progress and challenges,” Applied Physics Reviews 6, 021314 (2019), doi:10.1063/1.5088164.
• F. Diedrich, J. C. Bergquist, W. M. Itano, and D. J. Wineland, “Laser cooling to the zero-point energy of motion,” Physical Review Letters 62, 403–406 (1989), doi:10.1103/PhysRevLett.62.403.
• H. Häffner, C. F. Roos, and R. Blatt, “Quantum computing with trapped ions,” Physics Reports 469, 155–203 (2008), doi:10.1016/j.physrep.2008.09.003.
• D. F. V. James, “Quantum dynamics of cold trapped ions with application to quantum computation,” Applied Physics B 66, 181–190 (1998), doi:10.1007/s003400050373.
• D. Kielpinski, C. Monroe, and D. J. Wineland, “Architecture for a large-scale ion-trap quantum computer,” Nature 417, 709–711 (2002), doi:10.1038/nature00784.
• D. Leibfried, R. Blatt, C. Monroe, and D. Wineland, “Quantum dynamics of single trapped ions,” Reviews of Modern Physics 75, 281–324 (2003), doi:10.1103/RevModPhys.75.281.
• C. Monroe, D. M. Meekhof, B. E. King, S. R. Jefferts, W. M. Itano, D. J. Wineland, and P. Gould, “Resolved-sideband Raman cooling of a bound atom to the 3D zero-point energy,” Physical Review Letters 75, 4011–4014 (1995), doi:10.1103/PhysRevLett.75.4011.
• A. H. Myerson, D. J. Szwer, S. C. Webster, D. T. C. Allcock, M. J. Curtis, G. Imreh, J. A. Sherman, D. N. Stacey, A. M. Steane, and D. M. Lucas, “High-fidelity readout of trapped-ion qubits,” Physical Review Letters 100, 200502 (2008), doi:10.1103/PhysRevLett.100.200502.
• W. Paul, “Electromagnetic traps for charged and neutral particles,” Reviews of Modern Physics 62, 531–540 (1990), doi:10.1103/RevModPhys.62.531.
• D. J. Wineland and W. M. Itano, “Laser cooling of atoms,” Physical Review A 20, 1521–1540 (1979), doi:10.1103/PhysRevA.20.1521.
• D. J. Wineland, C. Monroe, W. M. Itano, D. Leibfried, B. E. King, and D. M. Meekhof, “Experimental issues in coherent quantum-state manipulation of trapped atomic ions,” Journal of Research of the National Institute of Standards and Technology 103, 259–328 (1998), doi:10.6028/jres.103.019.