Magneto-Optical Trap

Optical molasses can chill atoms almost to a standstill but it cannot hold them: three counter-propagating red-detuned beam pairs damp velocity beautifully, yet exert no force that depends on where an atom is, so a molasses cloud slowly diffuses apart. The magneto-optical trap fixes exactly this gap. It takes the velocity-dependent scattering force built in aq-optics-optical-molasses and aq-optics-doppler-cooling-limit and grafts onto it a position dependence, borrowed not from a new force but from an old piece of atomic structure — the Zeeman effect. A weak quadrupole magnetic field, vanishing at one point and growing linearly outward, tunes the atom's excited sublevels up on one side of centre and down on the other. Add circular polarizations of the right handedness and a displaced atom is pushed always toward the field zero, cooled as it goes. The ingredients are minimal: a pair of anti-Helmholtz coils makes a spherical quadrupole field $\mathbf{B}(\mathbf{r})$ ; six laser beams tuned a few linewidths below resonance supply the light; the atom's own $J_e > J_g$ level structure supplies the position sensitivity. Near the centre the net force is linear in velocity and displacement, $m\ddot{x} = -\beta\dot{x} - \kappa x$ — an overdamped harmonic oscillator made entirely of light. That single line is why the MOT is the workhorse of atomic physics: it grabs atoms from the room-temperature tail of a vapour and squeezes them into a cold, dense, well-localized ball. A MOT is deliberately a blunt instrument — its temperature is capped near the Doppler limit and its density by reabsorbed photons and light-assisted collisions. But as a fast, forgiving reservoir it is unmatched, routinely $10^{8}$ to $10^{10}$ atoms at tens of microkelvin: the mandatory first stage before the sub-Doppler cooling, magnetic or optical trapping, and evaporation that carry a gas to quantum degeneracy.

Key takeaways

A magneto-optical trap (MOT) crosses three pairs of counter-propagating, red-detuned, circularly polarized laser beams at the zero of a spherical quadrupole magnetic field to cool and confine neutral atoms at once.
The beams supply velocity damping (Doppler cooling / optical molasses); the inhomogeneous field makes the scattering force position-dependent through the Zeeman shift, adding a restoring force.
Getting the two circular handednesses right ( $\sigma^+$ vs $\sigma^-$ ) is what makes a displaced atom scatter preferentially from the beam pushing it home; get it wrong and the trap expels atoms.
Near centre the motion is an overdamped harmonic oscillator, $m\ddot{x}=-\beta\dot{x}-\kappa x$ , with the spring constant $\kappa$ and damping $\beta$ locked to the same slope of the scattering force, $\kappa/\beta = \mu' g B'/(\hbar k)$ .
A MOT captures $\sim 10^{9}$ atoms in seconds near the Doppler limit ( $\approx 146\ \mu$ K for $^{87}$ Rb), the standard gateway to magnetic trapping, evaporative cooling and Bose–Einstein condensation.
Key idea: the MOT superimposes two effects on one scattering force — red-detuned counter-propagating beams give velocity-dependent damping (cooling), and the Zeeman-shifted resonance in a quadrupole field makes the force position-dependent (trapping) — so with the right circular polarizations both collapse into $m\ddot{x}=-\beta\dot{x}-\kappa x$ .

The core insight

On one scattering force, red-detuned counter-propagating beams give velocity-dependent damping (cooling) while the Zeeman-shifted resonance in a quadrupole field makes it position-dependent (trapping), so with the right circular polarizations both collapse into the overdamped oscillator $m\ddot{x} = -\beta\dot{x} - \kappa x$ .

Where this lives today

Open almost any ultracold-atom laboratory and the first light you see is a MOT. It is the universal front end: it collects $10^{8}$ – $10^{10}$ atoms in seconds from a vapour or a Zeeman-slowed beam, then — after a brief sub-Doppler molasses stage and optical pumping — hands the cloud to a conservative magnetic or optical-dipole trap for evaporative cooling to Bose–Einstein condensation or a degenerate Fermi gas. That same switchable cold source underlies the most demanding instruments in physics: fountain and optical-lattice clocks reach fractional stabilities beyond $10^{-18}$ from MOT-loaded samples, and atom interferometers use MOT-prepared ensembles as inertial and gravitational sensors. The variants matter as much as the original. A dark-SPOT MOT masks the repump beam at the cloud centre, parking atoms in a dark state to suppress the reabsorbed-photon pressure and boost density tenfold; compressed-MOT ramps of detuning and gradient spike the phase-space density just before transfer. At the small end, a tightly focused MOT feeding optical tweezers loads single atoms through the collisional-blockade effect — the loading stage for the neutral-atom quantum computers and Rydberg arrays now assembling hundreds of addressable qubits. From the world's best clocks to programmable quantum simulators, the humble crossing of six beams and a pair of coils is where it all begins.

The physical picture

Start with optical molasses along one axis: two red-detuned beams from $\pm x$ . A moving atom sees the beam it runs into Doppler-shifted toward resonance and the one it runs away from shifted off, so it scatters more from the opposing beam — a friction force $-\beta v$ . That damps motion but says nothing about position; set the velocity to zero anywhere and the beams balance. Molasses is a viscous fluid of light with no walls. The MOT builds the walls out of the atom's internal structure. Superimpose a field that is zero at centre and grows linearly, $B = B'x$ . Through the Zeeman effect this splits the excited sublevels $m_e = -1, 0, +1$ in proportion to $x$ : at $x>0$ the $m_e=+1$ level rises and $m_e=-1$ falls, and at $x<0$ they swap. Give the beam from $-x$ helicity $\sigma^+$ ( $\Delta m = +1$ only) and the beam from $+x$ helicity $\sigma^-$ ( $\Delta m = -1$ ). At $x>0$ the falling $m_e=-1$ level slides toward the red-detuned laser, so the $\sigma^-$ beam — coming from $+x$ , pushing the atom back to centre — scatters more; displace the other way and the roles flip. The result is a restoring force $-\kappa x$ for no new physics: the field does no direct work, it merely tunes which beam wins.

Figure

MOT layout six red-detuned beams meet at the zero of an anti-Helmholtz quadrupole field with opposite circular polarizations σ+ and σ- on each axis, a radial gradient ≈ 10 G/cm and axial ≈ 20 G/cm, and a cloud of ~10⁹ atoms at the centre where B = 0

Layer the two effects and position (Zeeman) and velocity (Doppler) enter the resonance condition on exactly the same footing, through the single combination $\delta \mp kv \mp \mu' g_F B'x/\hbar$ . That is why one mechanism delivers both jobs, and why the damping $\beta$ and the spring $\kappa$ come out proportional. The whole cloud sits in an overdamped harmonic well of light: no potential energy is stored, so atoms do not oscillate — they creep to centre while bleeding kinetic energy into scattered photons. The next figure is the mechanism itself, an energy-versus-position diagram with the numbers a real $^{87}$ Rb trap runs at.

Figure

Mechanism of the MOT the excited sublevels m_e = +1, 0, −1 split linearly with position through the Zeeman shift with slope μ_B B′/h ≈ 1.4 MHz per mm at 10 G/cm while the red-detuned laser sits at δ = −2Γ ≈ −12 MHz below bare resonance; at x greater than 0 the falling m_e = −1 level meets the laser so the σ− beam scatters more and pushes the atom back toward centre

The field is genuinely weak — order $10\ \mathrm{G/cm}$ , a Zeeman shift of only $\approx 1.4\ \mathrm{MHz}$ per millimetre, with a direct magnetic force on a ground-state atom negligible even against gravity. Its whole contribution is to give light a position dependence: remove the light and the trap vanishes; remove the field and you are back to plain molasses.

The core equations

Everything begins with the photon scattering rate of a driven two-level atom, because the force is the momentum $\hbar k$ per absorption times that rate. For a single beam of saturation parameter $s_0 = I/I_{\rm sat}$ and effective detuning $\delta_{\rm eff}$ (which absorbs both the Doppler and Zeeman contributions), the steady-state force is a Lorentzian in $\delta_{\rm eff}$ .

F_{\mathrm{sc}}(\delta_{\rm eff}) = \hbar k\,\frac{\Gamma}{2}\,\frac{s_0}{1+s_0+\left(2\delta_{\rm eff}/\Gamma\right)^2} \qquad (1)

where: $\hbar k$ — photon momentum, $k = 2\pi/\lambda$ ; $\Gamma$ — natural linewidth; $s_0 = I/I_{\rm sat}$ — saturation parameter; $\delta_{\rm eff}$ — effective detuning including Doppler and Zeeman shifts. Physical reading: the force saturates at $F_{\max} = \hbar k\,\Gamma/2$ (at most half the population is excited) and peaks at $\delta_{\rm eff} = 0$ , resonance in the atom's own frame. The two beams of a pair carry position and velocity through their detunings. For an atom at displacement $x$ moving at $v$ , the Doppler shift $\mp kv$ and the Zeeman shift $\mp\mu' g_F B'x/\hbar$ enter with opposite signs for the two beams — that opposite sign is the whole game.

\delta_{\pm} = \delta \mp k v \mp \frac{\mu'\, g_F\, B'\, x}{\hbar} \qquad (2)

where: $\delta = \omega_L - \omega_0$ — laser detuning (red, $\delta < 0$ ); $kv$ — Doppler shift; $\mu' = (g_e m_e - g_g m_g)\mu_B$ — effective magnetic moment of the transition; $B' = \mathrm{d}B/\mathrm{d}x$ — field gradient; $x$ — displacement from the field zero. Physical reading: the $+$ beam ( $\sigma^+$ from $-x$ ) and $-$ beam ( $\sigma^-$ from $+x$ ) drive opposite $m$ sublevels, so a positive displacement pulls one toward resonance and the other away. Subtract the two single-beam forces and expand to first order in $v$ and $x$ . Because both enter only through $\delta_{\pm}$ , both come out multiplied by the same derivative of the Lorentzian, so the net force near centre is linear in each.

F_{\rm MOT} = F_{\rm sc}(\delta_+) - F_{\rm sc}(\delta_-) \approx -\beta v - \kappa x \qquad (3)

where: $\beta$ — velocity damping coefficient; $\kappa$ — restoring (spring) constant. Physical reading: this is the defining result — an atom near the centre of a MOT obeys a damped harmonic oscillator, cooling ( $\beta$ ) and trapping ( $\kappa$ ) delivered by one force. The damping is exactly the optical-molasses friction, from the velocity term alone, largest for red detuning near $\delta \approx -\Gamma/2$ . The spring constant is not independent — it inherits the same slope, scaled by the ratio of Zeeman shift per length to Doppler shift per velocity.

\beta = -\hbar k^2\,\frac{4 s_0\,(2\delta/\Gamma)}{\left[1+s_0+(2\delta/\Gamma)^2\right]^2},\qquad \kappa = \frac{\mu'\, g_F\, B'}{\hbar k}\,\beta \qquad (4)

where: $\delta < 0$ makes $\beta > 0$ (true damping); $s_0$ — per-beam saturation; the prefactor $\mu' g_F B'/(\hbar k)$ is a frequency converting Zeeman shift per length into an equivalent Doppler shift. Physical reading: detuning and intensity move $\beta$ and $\kappa$ together; only $B'$ changes their ratio, so — since $\kappa \propto \beta$ — the overdamping factor is fixed by parameters, not by turning up the light. The field itself is a spherical (anti-Helmholtz) quadrupole: divergence-free, with a single zero at the origin, and twice as steep along the coil axis as in the radial plane.

\mathbf{B}(\mathbf{r}) = B'\left(x\,\hat{\mathbf{x}} + y\,\hat{\mathbf{y}} - 2z\,\hat{\mathbf{z}}\right) \qquad (5)

where: $B'$ — radial gradient set by the coils; $\hat{\mathbf{z}}$ — coil axis, where the gradient is $-2B'$ . Physical reading: $\nabla\cdot\mathbf{B}=0$ forces the factor of two, so the axial direction is trapped twice as stiffly ( $\kappa_z = 2\kappa_{x,y}$ ) — real clouds are visibly compressed along the coil axis. The steady-state temperature balances recoil heating (a momentum random walk from randomly directed spontaneous emission) against damping, through $k_BT = D/\beta$ . Minimizing over detuning gives the Doppler limit.

k_B T = \frac{\hbar\Gamma}{4}\,\frac{1+s_0+(2\delta/\Gamma)^2}{2|\delta|/\Gamma}\;\xrightarrow[\ s_0\to 0\ ]{\ \delta=-\Gamma/2\ }\;k_B T_D = \frac{\hbar\Gamma}{2} \qquad (6)

where: $T$ — steady-state temperature; $T_D$ — Doppler limit, reached at $\delta = -\Gamma/2$ and vanishing intensity. Physical reading: the two-level MOT is no colder than molasses; real multilevel atoms beat this via polarization gradients, but the Doppler formula sets the scale and the optimal detuning. Finally, the cloud is a thermal Gaussian in the harmonic well, so size and peak density follow from equipartition.

x_{\rm rms} = \sqrt{\frac{k_B T}{\kappa}},\qquad n_0 = \frac{N}{(2\pi)^{3/2}\,x_{\rm rms}^3} \qquad (7)

where: $x_{\rm rms}$ — Gaussian radius; $N$ — atom number; $n_0$ — peak density. Physical reading: at fixed stiffness a colder cloud is smaller and denser — until reabsorbed photons (radiation trapping) push outward and cap density near $10^{11}\ \mathrm{cm^{-3}}$ .

Derivation AThe restoring force in the J=0 to J=1 model

The goal is to show, in the simplest transition that contains the physics, how circularly polarized beams plus a quadrupole field manufacture a force proportional to displacement, and why the resulting spring constant is welded to the molasses damping. We work one axis $x$ with a ground state $J_g = 0$ (no sublevels) and an excited state $J_e = 1$ with $m_e = -1, 0, +1$ ; the beam from $-x$ is $\sigma^+$ (drives $m_e=+1$ ) and the beam from $+x$ is $\sigma^-$ (drives $m_e=-1$ ). Everything hinges on which beam is nearer resonance when the atom is off-centre.

Step 1. Write the Zeeman shifts. In the field $B = B'x$ the sublevel $m_e$ shifts by $\Delta E_{m_e} = m_e\, g_e\mu_B B'x$ , so for $x>0$ (with $B'>0$ ) the $m_e=+1$ level rises and $m_e=-1$ falls; at $x<0$ they swap. Why: this is the only place the magnetic field enters — not as a force, but as a position-dependent tuning of the internal levels the light couples to.

Step 2. Fold the shifts into effective detunings: the two beams see $\delta_{\pm} = \delta \mp kv \mp g_e\mu_B B'x/\hbar$ . At $x>0$ the $\sigma^-$ beam (from $+x$ , driving the lowered $m_e=-1$ level) moves toward the red-detuned laser while $\sigma^+$ moves away. Why: the opposite-sign Zeeman terms, inherited from the opposite handedness, turn a symmetric molasses into a position-sensitive one.

Step 3. Form the force imbalance. The net force is $F = \hbar k\frac{\Gamma}{2}\big[\,s_0/(1+s_0+(2\delta_+/\Gamma)^2) - s_0/(1+s_0+(2\delta_-/\Gamma)^2)\,\big]$ , which for $x>0$ is negative (inward) because the $\sigma^-$ term dominates, and for $x<0$ is positive — a genuine restoring force, provided the polarizations match the sign of the gradient. Why: reverse either the handedness or the gradient sign and the imbalance flips, turning the trap into an anti-trap that blows atoms out.

Step 4. Linearize about centre. To first order $F \approx 2\,(\partial F_{\rm sc}/\partial\delta_{\rm eff})\,(kv + g_e\mu_B B'x/\hbar) = -\beta v - \kappa x$ , with $\kappa = (g_e\mu_B B'/\hbar k)\,\beta$ . Why: velocity and position ride the same derivative of the Lorentzian, so cooling and trapping are not independent — their ratio is fixed at $g_e\mu_B B'/(\hbar k)$ , and you cannot dial in trapping without cooling. *Verified: reproduces Foot, Atomic Physics eqs. (9.30)–(9.34) and Metcalf & van der Straten ch. 11; the fixed ratio $\kappa/\beta = \mu' g_F B'/(\hbar k)$ is standard.*

Go deeper · Derivation B · 2 minOverdamped motion and the cloud temperature

Derivation BOverdamped motion and the cloud temperature

The goal is to turn the linearized force into the trajectory of a real atom and the size of a real cloud, and to pin down honestly how overdamped the motion is — a point the textbook shorthand often overstates. Adding the recoil noise $f(t)$ from every scattered photon, one axis becomes a Langevin oscillator whose steady state fixes both temperature and cloud radius by the same balance that governs Brownian motion. We compare the rates, invoke equipartition, and read off the size.

Step 1. Write the equation of motion. With noise, the atom obeys $m\ddot{x} + \beta\dot{x} + \kappa x = f(t)$ , a damped oscillator driven by white momentum noise of strength set by the diffusion coefficient $D = (\hbar k)^2 R_{\rm sc}$ . Why: absorption gives a directed kick but spontaneous emission is randomly directed, so the same scattering that damps also heats — trapping and heating are two faces of one process.

Step 2. Compare the rates. The damping rate is $\gamma_{\rm damp} = \beta/m$ and the oscillation frequency $\omega_{\rm MOT} = \sqrt{\kappa/m}$ ; motion is overdamped when $\gamma_{\rm damp} > 2\omega_{\rm MOT}$ . For typical alkali parameters ( $s_0 \sim 1$ , $\delta \sim -\Gamma$ , $B' \sim 10\ \mathrm{G/cm}$ ) this ratio is only about $3$ — the trap is overdamped, but by a factor of order a few, not the $10^3$ – $10^4$ sometimes quoted. Why: since $\kappa \propto \beta$ , the ratio reduces to $\sqrt{\gamma_{\rm damp}\,\hbar k/(\mu' g_F B')}$ , a slowly varying combination — intensity cannot make the trap dramatically more overdamped; only a smaller gradient can.

Step 3. Apply equipartition. In steady state the harmonic coordinate carries $\tfrac12 k_B T$ despite the force being dissipative, because noise and damping balance: $\tfrac12\kappa\langle x^2\rangle = \tfrac12 k_B T$ , giving the Gaussian radius $x_{\rm rms} = \sqrt{k_B T/\kappa}$ . Why: even though no energy is stored in the spring, the equilibrium distribution of positions is identical to that of a real thermal oscillator — the cloud is a Boltzmann sample of a harmonic well.

Step 4. Fix the temperature. The Brownian steady state gives $k_B T = D/\beta$ , the same fluctuation–dissipation balance as molasses, so the two-level MOT temperature is the Doppler value $k_B T_D = \hbar\Gamma/2$ at optimal detuning. Why: the MOT adds trapping without changing the cooling physics, so it inherits molasses' temperature limit — beating it needs the multilevel, polarization-gradient mechanisms of sub-Doppler cooling. *Verified: consistent with the Metcalf & van der Straten Langevin treatment; the overdamping factor of order a few is confirmed numerically for $^{87}$ Rb at $s_0=1$ , $\delta=-\Gamma$ , $B'=10\ \mathrm{G/cm}$ ( $\gamma_{\rm damp}/\omega_{\rm MOT} \approx 3.1$ ).*

Limiting cases

Case	Result	Meaning
Small $v$ , small $x$ (linear regime)	$m\ddot{x} = -\beta\dot{x} - \kappa x$	Near centre the force is linear in both; overdamped harmonic motion defines the trap stiffness.
Doppler temperature limit	$k_B T_D = \hbar\Gamma/2$	At $\delta = -\Gamma/2$ , $s_0\to 0$ . For $^{87}$ Rb ( $\Gamma/2\pi = 6.07\ \mathrm{MHz}$ ), $T_D \approx 146\ \mu\mathrm{K}$ .
Recoil temperature floor	$k_B T_r = (\hbar k)^2/m$	Single-photon recoil sets the ultimate scale; $^{87}$ Rb gives $T_r \approx 362\ \mathrm{nK}$ . Real MOTs sit well above it.
Sub-Doppler regime	$T \ll T_D$	Multilevel atoms with polarization gradients reach tens of $\mu\mathrm{K}$ in and after the MOT, below the two-level prediction.
Axial vs radial stiffness	$\kappa_z = 2\,\kappa_{x,y}$	The quadrupole's factor of two makes the cloud narrower along the coil axis.
Saturated / dense regime	$F \to \hbar k\,\Gamma/2$	Each beam scatters near maximum; density is then capped near $10^{11}\ \mathrm{cm^{-3}}$ by radiation trapping and light-assisted collisions.

Common trapThe magnetic field is what holds the atoms in a MOT

It is not. The quadrupole gradient is weak (of order $10\ \mathrm{G/cm}$ ) and the direct magnetic force on a slow ground-state alkali atom is negligible — switch off the lasers and the cloud falls under gravity, trap gone. The field's only role is to Zeeman-shift the excited sublevels so that the light's scattering force becomes position-dependent; the confinement is entirely optical and dissipative, not magnetic and conservative. Two more traps to avoid: the polarization handedness must match the sign of the field gradient — reverse either and the "restoring" force becomes anti-restoring, expelling atoms — and the tidy 1D two-level picture explains the mechanism but not the true (often sub-Doppler) temperatures reached by real multilevel atoms.

Worked examples

1 · The Doppler limit and recoil floor for rubidium (confidence builder)

The $^{87}$ Rb cooling transition has $\Gamma/2\pi = 6.07\ \mathrm{MHz}$ at $\lambda = 780\ \mathrm{nm}$ , with atomic mass $m = 86.9\,u$ . Find the Doppler-limit temperature and the corresponding 1D rms speed, then the single-photon recoil temperature and recoil velocity.

Step 1. Doppler temperature. At optimal detuning $k_B T_D = \hbar\Gamma/2$ , with $\Gamma = 2\pi(6.07\times10^6) = 3.81\times10^7\ \mathrm{s^{-1}}$ , so $T_D = \hbar\Gamma/(2k_B) = (1.055\times10^{-34})(3.81\times10^7)/(2\times1.381\times10^{-23}) = \mathbf{1.46\times10^{-4}\ K \approx 146\ \mu K}$ .

Step 2. The rms speed at that temperature. Using $\tfrac12 m v_{\rm rms}^2 = \tfrac12 k_B T_D$ per axis, $v_{\rm rms} = \sqrt{k_B T_D/m} = \sqrt{(1.381\times10^{-23})(1.46\times10^{-4})/(1.443\times10^{-25})} = \mathbf{0.118\ m/s \approx 12\ cm/s}$ . Cold, but each atom is still moving at walking pace — which is why confinement, not just cooling, is needed.

Step 3. Recoil floor. One absorbed photon changes the velocity by $v_r = \hbar k/m = \mathbf{5.9\ mm/s}$ , and $k_B T_r = (\hbar k)^2/m$ gives $T_r = \mathbf{362\ nK}$ . The Doppler limit sits a factor $T_D/T_r \approx 400$ above this floor, a wide window that sub-Doppler cooling later exploits. T_D ≈ 146 μK, v_rms ≈ 12 cm/s, T_r ≈ 362 nK, v_r ≈ 5.9 mm/s.

2 · Capture velocity: why a MOT grabs only the slow tail

A single beam scatters at most $F_{\max} = \hbar k\,\Gamma/2$ , a deceleration $a_{\max}$ . For a trapping region of radius $r = 1\ \mathrm{cm}$ , estimate the fastest atom that can be stopped before it escapes, and compare with a room-temperature vapour.

Step 1. Maximum deceleration. $a_{\max} = \hbar k\,\Gamma/(2m) = \mathbf{1.12\times10^{5}\ m/s^2}$ , about $10^4$ times gravity — the scattering force is fierce, but acts only over the small beam overlap.

Step 2. Capture velocity. An atom entering at $v_c$ is stopped in distance $v_c^2/(2a_{\max})$ ; setting this equal to $r$ gives $v_c = \sqrt{2 a_{\max} r} = \sqrt{2(1.12\times10^5)(0.01)} = \mathbf{47\ m/s}$ . Only atoms slower than this can be turned around within the trap.

Step 3. Compare with the vapour. At $300\ \mathrm{K}$ the most probable Rb speed is $v_p = \sqrt{2k_B T/m} \approx 240\ \mathrm{m/s}$ , five times the capture velocity. Only the low-velocity tail — a fraction of order $(v_c/v_p)^3 \sim 10^{-2}$ — is captured directly, which is why a vapour-cell MOT still loads $\sim10^9$ atoms/s yet a Zeeman slower boosts that by orders of magnitude. v_capture ≈ 47 m/s, versus v_p ≈ 240 m/s — only the slow tail is trapped.

3 · Spring, damping, and cloud size of a working trap

Take $^{87}$ Rb with per-beam $s_0 = 1$ , detuning $\delta = -\Gamma$ , and radial gradient $B' = 10\ \mathrm{G/cm} = 0.10\ \mathrm{T/m}$ , with $\mu' \approx \mu_B$ . Find the damping rate, trap frequency, whether the motion is overdamped, and the rms cloud radius.

Step 1. Damping. From equation (4) with $2\delta/\Gamma = -2$ and $s_0 = 1$ , $\beta = \hbar k^2\cdot 4(1)(2)/(1+1+4)^2 = 0.222\,\hbar k^2 = 1.52\times10^{-21}\ \mathrm{kg/s}$ , so $\gamma_{\rm damp} = \beta/m = \mathbf{1.05\times10^{4}\ s^{-1}}$ (a damping time of $\approx 95\ \mu\mathrm{s}$ ).

Step 2. Spring and frequency. The Zeeman-to-Doppler prefactor is $\mu_B B'/(\hbar k) = (9.27\times10^{-24})(0.10)/(8.49\times10^{-28}) = 1.09\times10^{3}\ \mathrm{s^{-1}}$ , so $\kappa = (1.09\times10^3)\beta = 1.66\times10^{-18}\ \mathrm{N/m}$ and $\omega_{\rm MOT} = \sqrt{\kappa/m} = 3.4\times10^{3}\ \mathrm{rad/s}$ , i.e. $\mathbf{f_{\rm MOT} \approx 540\ Hz}$ .

Step 3. Overdamped? Compare $\gamma_{\rm damp}/\omega_{\rm MOT} = 1.05\times10^4/3.4\times10^3 = \mathbf{3.1} > 2$ , so the motion is overdamped — but only by a factor of a few, not the thousands sometimes claimed. A displaced atom relaxes to centre in $\sim95\ \mu\mathrm{s}$ , far shorter than an oscillation period ( $\sim1.9\ \mathrm{ms}$ ), so it never rings.

Step 4. Cloud size. Equation (6) here gives $T \approx 218\ \mu\mathrm{K}$ (above the $146\ \mu\mathrm{K}$ optimum, since $\delta=-\Gamma$ , $s_0=1$ is not the temperature-optimal point), so $x_{\rm rms} = \sqrt{k_B T/\kappa} = \mathbf{43\ \mu m}$ . γ_damp ≈ 1.0×10⁴ s⁻¹, f_MOT ≈ 540 Hz, overdamped ×3, x_rms ≈ 43 μm.

Physics of the portrait

The MOT is the moment laser cooling grew a container. Molasses could slow atoms; trapping them took a conceptual sleight of hand — using the atom's own Zeeman structure to make a dissipative force behave like a restoring one — and a group that had spent a decade building the tools.

Origins · 1 minDavid E. Pritchard and how this was discovered

David E. Pritchard

born 1941 · neutral-atom trapping and precision atomic physics

David Pritchard, a physicist at MIT, is the figure most directly responsible for the idea of trapping neutral atoms with light. In the early 1980s, when laser cooling could slow atoms but not hold them, Pritchard analyzed atom traps in general and argued that the spontaneous scattering force — long thought too weak and, by an optical Earnshaw theorem, unable to make a stable trap alone — could trap after all if one broke the required symmetry using internal-state structure. The specific recipe — a quadrupole field plus opposite circular polarizations, so the Zeeman shift supplies the position dependence — was proposed by Jean Dalibard and realized in 1987 by Pritchard, Steven Chu and their coworkers Raab, Prentiss and Cable, using sodium in six beams and anti-Helmholtz coils. It was an immediate revolution: robust, forgiving, loading straight from a vapour. Pritchard's MIT group became a wellspring of the field — his students include Wolfgang Ketterle, who used exactly this load-then-evaporate strategy for the first sodium condensate, and Eric Cornell. The MOT remains, four decades on, the first thing nearly every cold-atom apparatus switches on.

History

1975

— Hänsch and Schawlow propose Doppler cooling of neutral atoms with counter-propagating laser beams

1985

— Chu and coworkers at Bell Labs demonstrate optical molasses in three orthogonal beam pairs

1987

— Raab, Prentiss, Cable, Chu and Pritchard build the first magneto-optical trap, on a proposal by Dalibard

1988

— Lett, Phillips and coworkers measure sub-Doppler temperatures, soon explained by the Sisyphus mechanism

1997

— Chu, Cohen-Tannoudji and Phillips share the Nobel Prize for laser cooling and trapping of atoms

Going further

Texts. Foot, Atomic Physics, ch. 9 (laser cooling and trapping), eqs. 9.30–9.34 · Metcalf & van der Straten, Laser Cooling and Trapping (Springer, 1999), ch. 11 · Demtröder, Laser Spectroscopy, Vol. 2, ch. on cooling and trapping · Saleh & Teich, Fundamentals of Photonics, ch. on atom optics · Steck, Alkali D Line Data (rubidium, cesium) for linewidths, wavelengths and saturation intensities.

Provenance. The mechanism follows the standard two-level $J_g=0\to J_e=1$ model of Foot ch. 9 and Metcalf & van der Straten ch. 11; the first MOT is Raab, Prentiss, Cable, Chu & Pritchard, Phys. Rev. Lett. 59, 2631 (1987), on a proposal by J. Dalibard. A verifier should confirm: (i) $F_{\max}=\hbar k\,\Gamma/2$ and the fixed ratio $\kappa/\beta = \mu' g_F B'/(\hbar k)$ ; (ii) the Doppler limit $k_BT_D=\hbar\Gamma/2$ giving $T_D\approx146\ \mu$ K for $^{87}$ Rb ( $\Gamma/2\pi=6.07$ MHz); (iii) recoil $T_r=(\hbar k)^2/(mk_B)\approx362$ nK, $v_r=\hbar k/m\approx5.9$ mm/s; (iv) the divergence-free quadrupole with axial:radial gradient 2:1; (v) the trap numbers $\gamma_{\rm damp}\approx1.05\times10^4\ \mathrm{s^{-1}}$ , $f_{\rm MOT}\approx540$ Hz, overdamping $\approx3.1$ , $x_{\rm rms}\approx43\ \mu$ m at $s_0=1$ , $\delta=-\Gamma$ , $B'=10$ G/cm. Note: the overdamping factor is of order a few, not $10^3$ – $10^4$ ; all other numbers are standard textbook/database values.

Prerequisites. Radiation Pressure and the Scattering Force · Doppler Cooling and the Doppler Limit · Optical Molasses. Leads to. Sub-Doppler and Sisyphus Cooling · Magnetic Trapping of Neutral Atoms · Evaporative Cooling · Bose–Einstein Condensation · Zeeman Slower.

Ask the tutor: Ask me to walk through how the $\sigma^+/\sigma^-$ handedness and the sign of $B'$ conspire to make the force restoring — and what physically happens if you flip just one of them.

Is the MOT a magnetic trap for the atoms?

No, and the name misleads. The gradient is weak (order $10\ \mathrm{G/cm}$ ) and its direct force on a slow ground-state atom is negligible. Its sole job is to Zeeman-shift the excited sublevels so the atom's resonance condition depends on position; the light's scattering force then acquires a restoring component. Switch off the lasers and the trap vanishes instantly — the atoms simply fall. A genuine magnetic trap, capturing atoms by $-\nabla(\boldsymbol{\mu}\cdot\mathbf{B})$ on their moment, is what a MOT usually loads into afterwards, with steeper fields and no light.

Why does the trap need a J_e greater than J_g transition?

Because the position dependence comes entirely from Zeeman-shifting the excited sublevels, and you need sublevels to shift. In the model transition $J_g=0\to J_e=1$ the ground state is a single level while the excited state splits into $m_e=-1,0,+1$ ; the field moves these apart with displacement, and the $\sigma^+$ and $\sigma^-$ beams each address a different one. With no excited substructure ( $J_e=0$ ) there is nothing for the field to shift, the force stays position-independent, and you have plain molasses — cooling but no confinement.

How is the spring constant related to the damping coefficient?

They are proportional, not independent: $\kappa = (\mu' g_F B'/\hbar k)\,\beta$ . Both come from the same slope of the scattering force versus effective detuning, because velocity and position enter that detuning identically. The extra factor in $\kappa$ just converts a Zeeman shift per unit length into an equivalent Doppler shift per unit velocity. So changing intensity or detuning moves cooling and trapping together — the only knob that changes their ratio, and hence how overdamped the trap is, is the field gradient $B'$ .

Why does the cloud look squashed along one axis?

Because the quadrupole is stiffer along the coil axis. Maxwell's law $\nabla\cdot\mathbf{B}=0$ forces $\mathbf{B}=B'(x\hat{\mathbf{x}}+y\hat{\mathbf{y}}-2z\hat{\mathbf{z}})$ , so the axial gradient is twice the radial one. Since $\kappa$ scales with the gradient, $\kappa_z=2\kappa_{x,y}$ , and $x_{\rm rms}\propto\sqrt{1/\kappa}$ , the axial radius is smaller by $\sqrt{2}$ . A MOT cloud is therefore an ellipsoid compressed along the field axis, not a sphere — a handy sign that coils and beams are aligned to the field zero.

What limits the temperature and density of a simple MOT?

Two separate bottlenecks. Temperature, in the two-level picture, is the Doppler limit $k_BT_D=\hbar\Gamma/2$ — recoil heating balancing velocity damping — though multilevel atoms slip below it via polarization-gradient cooling. Density is capped by radiation trapping (a photon scattered by one atom is reabsorbed by a neighbour, giving an outward push) and by light-assisted collisions that eject atom pairs, holding the peak near $10^{11}\ \mathrm{cm^{-3}}$ . Dark-SPOT MOTs beat the radiation-trapping limit by hiding central atoms in a dark state.

Is the MOT force conservative, and why does that matter?

No — and this is its defining feature. The trapping force $-\kappa x$ comes from photon scattering, a dissipative process, so it is not $-\nabla U$ . That is exactly what lets the MOT cool and trap at once: a conservative trap holds atoms but never removes their energy, whereas the MOT continuously bleeds kinetic energy into scattered photons while confining. The flip side is that it cannot preserve coherence and its temperature is floored by scattering — which is why every experiment eventually transfers atoms out into a conservative trap, kills the light, and evaporates.