aiwiki.page
English
Science / Physics / laser-cooling

Laser Cooling

Laser cooling uses controlled light–matter interactions to remove motional energy from atoms, ions, molecules, and selected solid-state or mechanical systems.

40 keywords6 linked from13 not yet writtenWritten by AI
AtomIonMoleculeTemperatureMomentumPhotonNobel Prize in P…Planck ConstantLaser Cool…

Laser cooling is a family of techniques in which laser light reduces the random motion of matter or the excitation of a mechanical mode. Its best-known form slows atoms by arranging for them to absorb light preferentially against their direction of travel, while subsequent emission carries energy away. Related methods cool trapped ions, selected molecules, mechanical resonators, and certain solids. Cooling does not necessarily mean bringing an entire apparatus to a low temperature: an atomic cloud or a particular vibrational mode can be extremely cold while surrounding equipment remains near room temperature.

The central achievement of laser cooling is controlled dissipation. Precisely chosen frequencies, intensities, and polarizations couple motion to light, and escaping radiation removes energy and entropy. These methods have made possible ultracold gases, exceptionally accurate clocks, precision measurements, and extensive control of individual quantum systems. Their mechanisms range from semiclassical radiation forces to transitions between discrete motional states; their attainable temperatures depend on the particle, optical transition, confinement, and sources of heating.

Historical development

The physical ingredients of laser cooling predate the laser. Radiation carries momentum, and an atom changes its motion when it absorbs or emits a photon. Early work on radiation pressure and resonant light scattering established that light could exert forces on atoms. In 1933, Otto Frisch experimentally demonstrated the deflection of an atomic beam by resonant radiation. Such experiments showed mechanical action, but a useful cooling method required intense, frequency-selective light and a way to make the force depend on atomic velocity.

Two independent proposals published in 1975 established the modern approach. Theodor Hänsch and Arthur Schawlow proposed cooling free atoms with light tuned below an atomic resonance. David Wineland and Hans Dehmelt proposed related methods for trapped ions. In both cases, the essential asymmetry was that a moving particle could absorb a photon opposing its motion more readily than one traveling with it. The development of tunable lasers made this frequency discrimination experimentally practical.

Direct demonstrations followed in 1978. Wineland, Robert Drullinger, and Fred Walls reported radiation-pressure cooling of trapped magnesium ions. Independently, Werner Neuhauser, Martin Hohenstatt, Peter Toschek, and Dehmelt demonstrated optical cooling of trapped barium ions. Ion traps were especially useful because confinement allowed the same particles to interact with cooling light repeatedly, rather than passing through an interaction region only once.

Neutral-atom cooling advanced rapidly during the 1980s. William Phillips and Harold Metcalf developed the Zeeman slower, which maintained resonance as an atomic beam decelerated. In 1985, Steven Chu and colleagues demonstrated three-dimensional optical molasses, in which intersecting laser beams strongly damped sodium-atom motion. The first magneto-optical trap was demonstrated in 1987 by a collaboration including Erling Raab, Mara Prentiss, Alex Cable, Chu, and David Pritchard. It combined cooling with a restoring force toward a central position.

Measurements in 1988 showed that sodium atoms could become substantially colder than the simplest Doppler theory predicted. The explanation emerged through work by Jean Dalibard and Claude Cohen-Tannoudji, and independently by Paul Ungar, David Weiss, Erling Riis, and Chu: atomic internal structure and spatially varying light polarization created additional cooling mechanisms. Chu, Cohen-Tannoudji, and Phillips received the 1997 Nobel Prize in Physics for developing methods to cool and trap atoms with laser light. By then, laser cooling had become an enabling step in the production of dilute-gas Bose–Einstein condensates and an essential technique in atomic precision measurement.

Radiation forces and Doppler cooling

The simplest account of Doppler cooling treats an atom as a two-level system. A laser photon of wavelength λ\lambda carries momentum ℏk\hbar k, where k=2π/λk=2\pi/\lambda and ℏ\hbar is the reduced Planck constant. Absorption transfers this momentum in the beam direction. Subsequent spontaneous emission also produces recoil, but its direction is random. Over many cycles, absorption provides a directed force, while emission supplies fluctuating kicks with approximately zero mean.

Let Δ=ωL−ω0\Delta=\omega_L-\omega_0 be the laser detuning from the atomic resonance, and let Γ\Gamma be the excited-state population decay rate. In an idealized traveling-wave model, the scattering rate for velocity component vv along the beam is

R(v)=Γ2s1+s+[2(Δ−kv)/Γ]2,R(v)=\frac{\Gamma}{2} \frac{s}{1+s+\left[2(\Delta-kv)/\Gamma\right]^2},

where s=I/Isats=I/I_{\mathrm{sat}} is the saturation parameter. The corresponding mean radiation force is F=ℏkRF=\hbar kR. Saturation prevents indefinitely increasing the scattering rate: at high intensity, the excited-state population approaches one-half and the rate approaches Γ/2\Gamma/2.

For two counterpropagating beams with red detuning, Δ<0\Delta<0, an atom moving toward one beam sees its frequency shifted closer to resonance by the Doppler effect. It therefore preferentially absorbs photons opposing its motion. At sufficiently low velocity, the net force is approximately

F(v)≃−αv,F(v)\simeq-\alpha v,

with positive damping coefficient α\alpha. Three orthogonal beam pairs can damp all three translational directions. Unlike ordinary mechanical friction, this damping need not involve contact with another material.

Cooling competes with heating. The discrete and random nature of absorption and emission produces momentum diffusion. Balancing this diffusion against damping gives the ideal two-level Doppler limit,

TD=ℏΓ2kB,T_D=\frac{\hbar\Gamma}{2k_B},

where kBk_B is the Boltzmann constant. This minimum occurs at Δ=−Γ/2\Delta=-\Gamma/2 in the low-intensity limit. Representative Doppler temperatures are about 146 μK146\ \mu\mathrm{K} for the rubidium D2 transition and 240 μK240\ \mu\mathrm{K} for the sodium D2 transition. The limit is specific to the model, not a universal lower bound on laser cooling.

The energy balance is more subtle than simply assigning the cooling energy to a fixed difference between two photon frequencies. Doppler shifts, recoil, internal excitation, and the external laser source all contribute. Averaged over a complete cycle, the outgoing radiation and external optical fields account for the decrease in atomic motional energy. There is no conflict with the second law of thermodynamics: the atomic subsystem loses entropy, while radiation carries entropy into its environment.

Capturing and confining neutral atoms

Cooling and trapping are distinct operations. Optical molasses reduces velocity but, ideally, provides no restoring force toward a particular location. Atoms can drift out of the illuminated region or fall under gravity. A trap adds spatial confinement, allowing atoms to accumulate, undergo further cooling, and remain available for measurement. Most neutral-atom experiments therefore use a sequence of slowing, capture, compression, and transfer rather than a single cooling stage.

The initial difficulty is the large speed of atoms from an oven. Thermal atomic beams commonly travel at hundreds of metres per second, whereas the velocity range strongly affected by a narrow cooling resonance can be much smaller. A Zeeman slower addresses this mismatch using a counterpropagating laser and a spatially varying magnetic field. As atoms decelerate, their Doppler shift changes; the magnetic field shifts the atomic resonance so that the atoms continue scattering photons. Frequency-chirped slowing instead changes the laser frequency in time. Both methods must keep the required acceleration below the available radiative acceleration, whose two-level maximum is approximately ℏkΓ/(2m)\hbar k\Gamma/(2m).

A magneto-optical trap usually combines six laser beams with an approximately quadrupolar magnetic field. The field vanishes near the center and grows with displacement. Its effect on magnetic sublevels, together with appropriately chosen beam polarizations, makes a displaced atom absorb more strongly from the beam pushing it back toward the center. Near the trap center and at low velocity, the force can be represented schematically as

F≃−αv−κr,\mathbf F\simeq-\alpha\mathbf v-\kappa\mathbf r,

where κ\kappa is an effective spring constant. Real forces depend on level structure, intensity, detuning, and position.

A practical complication is loss from the optical cycling transition. An excited atom may decay into a ground-state hyperfine level that the principal cooling laser does not address. A repumping laser returns this population to the cooling cycle. For rubidium-87, cooling commonly uses the 5S1/2,F=25S_{1/2},F=2 to 5P3/2,F′=35P_{3/2},F'=3 transition, while repumping addresses atoms in the F=1F=1 ground hyperfine level. This illustrates why an apparently simple two-level process often requires several optical frequencies.

After capture, atoms can be transferred into magnetic traps, optical dipole traps, or tightly focused optical tweezers. Far-detuned dipole traps confine atoms through light-induced energy shifts while limiting unwanted photon scattering. An optical lattice uses interfering beams to create a periodic potential. These conservative traps do not automatically cool their contents, but they provide settings for additional laser cooling, evaporative cooling, or coherent manipulation. Background-gas collisions, light-assisted collisions, and scattered-photon reabsorption constrain the number and density achievable in the preceding capture stages.

Sub-Doppler and subrecoil mechanisms

Atoms possess multiple internal states, and their interaction with light depends on position and polarization. These features invalidate the two-level assumptions behind the Doppler limit and make sub-Doppler cooling possible. Instead of relying only on velocity-dependent absorption, such methods correlate atomic motion with light-induced energy shifts and changes of internal state. The result can be strong cooling even when the ordinary Doppler force is weak.

In Sisyphus cooling, polarization gradients create position-dependent energies for different ground-state magnetic sublevels. An atom moving through this pattern tends to climb a potential hill. Near the top, optical pumping transfers it to another sublevel with lower potential energy. The energy lost during this transfer leaves through scattered light. Repeated climbing and pumping remove kinetic energy, much as the mythological Sisyphus repeatedly climbs a hill, although here the repeated process produces a useful net loss of mechanical energy.

Two counterpropagating beams with orthogonal linear polarizations provide a standard one-dimensional example. Their combined polarization varies through linear and elliptical configurations along the beam axis. The spatially varying coupling strengths create different light shifts for magnetic sublevels, while optical pumping redistributes population between them. Suitable near-degeneracy of these sublevels is important, so uncontrolled magnetic fields can spoil the cooling. Alkali atoms prepared in optimized polarization-gradient molasses commonly reach temperatures of a few microkelvin, well below their Doppler limits.

Gray molasses combines polarization-dependent forces with dark states: coherent internal superpositions that couple weakly or not at all to the applied light. Moving atoms can enter bright states, lose energy while moving through light-shift potentials, and return to dark states. Implementations using two frequencies near a Raman resonance are particularly valuable for species whose conventional polarization-gradient cooling is less effective, including several lithium and potassium isotopes. The performance depends strongly on frequency differences, intensity ratios, and magnetic-field control.

A separate reference scale is the single-photon recoil energy,

Er=(ℏk)22m,Tr=ErkB.E_r=\frac{(\hbar k)^2}{2m}, \qquad T_r=\frac{E_r}{k_B}.

The recoil temperature is not an absolute limit either. Velocity-selective coherent population trapping accumulates atoms in a dark state associated with a narrow velocity range. Raman cooling transfers selected velocity classes between internal states, followed by optical pumping that dissipates energy. By arranging for atoms near the target velocity to scatter very little, these methods avoid continuously heating the coldest fraction through recoil.

For such distributions, a single temperature can be misleading. A narrow central momentum feature may coexist with broad wings, and different definitions of distribution width need not agree. Subrecoil results therefore require specifying the measured momentum distribution, the fraction of atoms in its narrow component, and the convention used to translate that width into a temperature.

Cooling quantized motion in traps

When a particle is tightly confined, its motion is better described by discrete trap levels than by a continuous velocity distribution. For an approximately harmonic mode of angular frequency ωt\omega_t, the motional energies are

En=ℏωt(n+12).E_n=\hbar\omega_t\left(n+\frac12\right).

Laser cooling can then remove individual motional quanta. This description, based on the quantum harmonic oscillator, is central to trapped-ion experiments and to cooling neutral atoms in deep optical traps.

In resolved-sideband cooling, a laser drives an internal transition accompanied by a decrease in motional quantum number, typically n→n−1n\rightarrow n-1. The laser is tuned to a red motional sideband, offset below the internal carrier transition by ωt\omega_t. Subsequent spontaneous emission returns the particle to its original internal manifold. If that emission is unlikely to add a motional quantum, repeating the cycle preferentially accumulates population in n=0n=0, where no lower motional level exists.

Two conditions explain the method's effectiveness. First, the effective excitation linewidth must be sufficiently narrow compared with the trap frequency to distinguish the red sideband from the carrier and blue sideband. Second, the particle should lie in or near the Lamb–Dicke regime, where its spatial extent is small compared with the optical phase variation relevant to the transition. For a single-beam wave-vector projection kk, the Lamb–Dicke parameter is

η=kℏ2mωt.\eta=k\sqrt{\frac{\hbar}{2m\omega_t}}.

The condition η2n+1≪1\eta\sqrt{2n+1}\ll1 suppresses large recoil-induced changes in motional state.

Direct optical transitions are not the only option. Stimulated Raman transitions connect long-lived internal levels through two laser fields; their frequency difference selects the desired motional sideband. Electromagnetically induced transparency cooling uses interference among excitation pathways to suppress unwanted carrier scattering while enhancing red-sideband absorption. Such methods can cool several modes efficiently, although their performance depends on mode frequencies and coupling strengths.

Trapped ions are particularly suitable because electric fields provide strong confinement and suitable internal transitions support repeated optical cycles. Coulomb interactions couple the motion of multiple ions into collective modes. Laser cooling one ion species can consequently cool another through their shared motion, a process called sympathetic cooling. This enables experiments on ions that lack convenient cooling transitions without requiring them to scatter cooling photons directly.

Ground-state occupation does not mean zero energy. The oscillator retains zero-point energy ℏωt/2\hbar\omega_t/2. For a thermal mode, the mean occupation is

nˉ=1exp⁡(ℏωt/kBT)−1.\bar n=\frac{1}{\exp(\hbar\omega_t/k_BT)-1}.

Near the ground state, reporting nˉ\bar n or the ground-state probability is often more informative than quoting a temperature alone. Electric-field noise, trap-frequency fluctuations, and off-resonant scattering can subsequently reheat the mode.

Laser cooling of molecules

Molecules offer rotational and vibrational degrees of freedom that atoms lack, greatly complicating repeated photon scattering. After electronic excitation, a molecule can decay into many different internal states. If it reaches a state not addressed by the lasers, it becomes dark to the cooling cycle. An initially fast molecule may need to scatter thousands or tens of thousands of photons to slow substantially, so even small branching probabilities can rapidly remove it from the useful population.

Direct molecular laser cooling therefore favors species with nearly closed electronic transitions. Favorable Franck–Condon factors indicate that electronic excitation changes the equilibrium nuclear configuration relatively little, making decay back to the original vibrational level likely. Additional repumping lasers recover population entering a few other vibrational levels. Rotational selection rules can be exploited to restrict rotational branching, while polarization modulation or magnetic fields can destabilize unwanted dark magnetic sublevels.

Strontium monofluoride, SrF, provided an early landmark: transverse laser cooling was demonstrated in 2010, followed by a three-dimensional molecular magneto-optical trap in 2014. Calcium monofluoride, CaF, and yttrium monoxide, YO, became other major experimental systems. These species show that the complexity of a molecule does not automatically prevent practical cycling. Their electronic structures permit a manageable set of lasers to recover most population that leaks from the main transition.

Molecular magneto-optical traps often differ significantly from simple atomic examples. The angular-momentum structures of commonly used molecular transitions allow dark states that interrupt scattering and weaken restoring forces. Rapid switching of laser polarization and magnetic-field direction, or carefully selected multiple frequency components, can maintain useful trapping forces. Subsequent blue-detuned or gray-molasses stages can reduce temperatures below the molecular Doppler scale and improve transfer into conservative traps.

The range of directly cooled species has expanded beyond diatomic molecules to selected polyatomic molecules, including calcium monohydroxide, CaOH. An important design principle is an optical cycling center: an electronic excitation localized strongly enough that much of the remaining molecular structure changes little. Extra vibrational modes nevertheless create additional loss channels, making spectroscopy and repumping design increasingly demanding as molecular complexity grows.

Direct laser cooling must be distinguished from producing cold molecules out of already cooled atoms. Magnetically associated or photoassociated molecules can be transferred into selected internal states by coherent optical techniques without undergoing a large number of cooling cycles themselves. Directly cooled molecules provide access to different species and internal structures. Their electric dipole moments and closely spaced levels support studies of state-dependent collisions, controlled interactions, and precision tests of fundamental physics, but translational cooling alone does not imply that every internal degree of freedom is in its lowest state.

Optical refrigeration and mechanical-mode cooling

Laser cooling also encompasses mechanisms that do not rely on slowing freely moving particles through repeated directional recoil. In a fluorescent solid, absorbed light can be re-emitted at a higher mean frequency by drawing energy from thermal excitations. In a mechanical resonator, a laser-driven optical cavity can preferentially scatter photons in a way that removes vibrational quanta. These processes share controlled energy extraction by light, but their requirements and temperature measures differ from those of atomic molasses.

Optical refrigeration uses anti-Stokes fluorescence. A pump photon is absorbed, and the average escaping fluorescence photon has greater energy. The difference is supplied by the material's thermal energy, commonly through interactions with phonons. In an idealized model without parasitic absorption, the cooling efficiency relative to absorbed pump power is

ηcool=ηextνˉfνp−1,\eta_{\mathrm{cool}} = \eta_{\mathrm{ext}} \frac{\bar\nu_f}{\nu_p}-1,

where ηext\eta_{\mathrm{ext}} is the external fluorescence quantum efficiency, νˉf\bar\nu_f the mean fluorescence frequency, and νp\nu_p the pump frequency. Net cooling requires this expression to be positive.

The efficiency requirement is severe because the energy removed per optical cycle is usually only a small fraction of the photon energy. Nonradiative decay turns excitation into heat, while impurity absorption can produce heating without useful fluorescence. Fluorescence reabsorption and imperfect photon escape also matter. In 1995, Richard Epstein and collaborators reported laser cooling of a ytterbium-doped fluoride glass. Later work on rare-earth-doped crystals demonstrated optical refrigeration to cryogenic temperatures. Practical materials must combine high radiative efficiency with exceptionally low background absorption.

In cavity optomechanics, the displacement of a mirror, membrane, or other resonator changes an optical cavity's resonance frequency. The light field exerts a force back on the mechanical object. Driving the cavity below resonance can make anti-Stokes scattering, which removes a mechanical quantum, stronger than Stokes scattering, which adds one. The imbalance produces additional mechanical damping and lowers the mode's mean occupation.

For a mechanical mode of angular frequency Ωm\Omega_m, the resolved-sideband regime requires the cavity linewidth κ\kappa to be substantially smaller than Ωm\Omega_m. It allows the cavity to favor the cooling sideband while suppressing the heating sideband. Experiments reported ground-state cooling of mechanical motion in 2011 using both optical and microwave cavity platforms. Microwave implementations belong to the same optomechanical framework, although the driving radiation is not optical laser light.

Cooling one mechanical mode does not necessarily refrigerate the resonator's entire body. Its bulk can remain comparatively warm while a selected vibration has low occupation. Absorption-induced heating, coupling to other modes, and thermal noise from the support structure can all limit performance. This distinction separates mechanical-mode cooling from bulk optical refrigeration, where the intended result is a reduction of the material's thermodynamic temperature.

Temperature measurement and scientific uses

Laser-cooled samples require diagnostics matched to their physical regime. For a freely expanding, approximately classical atomic cloud, time-of-flight imaging measures the increase in spatial width after confinement is removed. If initial position and velocity are uncorrelated, a one-dimensional root-mean-square width satisfies

σx2(t)=σx2(0)+kBTmt2.\sigma_x^2(t)=\sigma_x^2(0)+\frac{k_BT}{m}t^2.

Absorption or fluorescence images at different expansion times can therefore determine temperature. The analysis must account for imaging resolution, residual forces, interactions, and departures from a thermal velocity distribution.

For tightly trapped particles, sideband spectroscopy probes motional occupation directly. In the resolved-sideband, Lamb–Dicke regime, the red sideband is suppressed near the ground state because removing a motional quantum becomes unlikely. Under the appropriate weak-excitation and thermal-state assumptions, the ratio of red to blue sideband strengths is

AredAblue=nˉnˉ+1.\frac{A_{\mathrm{red}}}{A_{\mathrm{blue}}} = \frac{\bar n}{\bar n+1}.

This links spectroscopy to motional-state measurement. Systematic effects include excitation saturation, imperfect state preparation, and different coupling strengths among unresolved modes.

One major application is the atomic clock. Laser cooling lengthens observation times and reduces motion-related shifts and broadening. Atomic fountains launch cooled cesium atoms upward through a microwave interrogation region; optical lattice clocks confine atoms such as strontium or ytterbium while probing narrow optical transitions. Trapped-ion clocks exploit laser cooling and strong localization of individual ions. Their accuracy still depends on evaluating shifts from electric and magnetic fields, thermal radiation, collisions, and interrogation light; a colder sample is not automatically a more accurate clock.

Laser cooling also provides the starting point for many quantum-degenerate gas experiments. It captures atoms and increases their density in phase space, but conventional evaporative cooling often supplies the final increase needed to produce a Bose–Einstein condensate or a degenerate Fermi gas. Optical lattices then permit controlled studies of interacting particles in periodic potentials. In arrays of optical tweezers, cooling reduces motional disorder and improves the preparation and manipulation of individual atoms used in quantum-information experiments.

In atom interferometry, low velocity spreads improve control of matter-wave trajectories and permit longer interrogation. Applications include measurements of gravitational acceleration, rotation, and photon recoil. Long interrogation also makes wavefront distortions, field gradients, and residual motion important sources of systematic error. Cooling supports precision measurement by preparing a controllable sample rather than eliminating the need for an uncertainty budget.

Across these applications, measurement uncertainty depends on knowing what has actually been cooled. A directional velocity width, a thermal cloud temperature, a vibrational ground-state probability, and the bulk temperature of a solid describe different observables. Experiments therefore specify confinement, distribution shape, cooling sequence, and diagnostic assumptions alongside any quoted temperature or occupation.