Sympathetic Ground-State Cooling

• Sympathetic ground-state cooling is a method where a target system is cooled indirectly via a coupled, directly dissipated refrigerant, achieving near-zero motional quanta.
• Techniques like resolved sideband, Raman, and electromagnetically induced transparency cooling are employed to reduce the mean phonon number (n ≪ 1) in systems ranging from trapped ions to hybrid setups.
• Diverse coupling mechanisms, from Coulombic interactions in ion crystals to optomechanical and electromechanical methods, enable practical applications in high-precision spectroscopy and quantum metrology.

Sympathetic ground-state cooling is the preparation of a target particle or oscillator near the quantum ground state of motion by cooling a different, coupled system that can be directly dissipated. In trapped-ion implementations, the coupling is usually Coulombic and the cooled object is a shared normal mode of an ion crystal; in hybrid mechanical and electromechanical platforms, the coupling can be optical, cavity-mediated, or electromechanical. Across these settings, the central observable is the mean phonon number $\bar n$, with near-ground-state operation corresponding to $\bar n \ll 1$, and the technique serves as a starting point for quantum-logic spectroscopy, precision metrology, optical clocks, and controlled cold-collision experiments (Rugango et al., 2014, Wan et al., 2015, Chen et al., 2016, Malossi et al., 2020).

1. Definition and conceptual boundaries

Sympathetic cooling differs from direct cooling in that the target does not need to scatter the cooling photons. In trapped-ion systems this is decisive for molecular ions, highly charged ions, and clock ions, which often lack strong closed optical cycling transitions. A coolant ion is laser-cooled, and because the motional degrees of freedom are shared through the Coulomb interaction, energy is removed from the collective modes that include the target species (Rugango et al., 2014, King et al., 2021).

Ground-state cooling is stricter than merely reaching the Doppler regime or forming a Coulomb crystal. The axial motion of a trapped ion is modeled as a quantum harmonic oscillator with energies $E_n=\hbar\omega(n+\tfrac12)$, and the thermal occupation is related to temperature by

$\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$

In this language, “near ground state” means $\bar n \ll 1$, whereas “translationally cold” in millikelvin-scale Coulomb crystals can still correspond to many motional quanta (Rugango et al., 2014, Tong et al., 2010).

The literature also distinguishes external from internal cooling. In sympathetically cooled molecular ions, the Coulomb interaction damps translational motion but does not efficiently couple to rotational or vibrational motion. This is why state-selective preparation of internal molecular states can be preserved during sympathetic translational cooling, and why ground-state cooling usually refers first to external motion unless explicitly stated otherwise (Tong et al., 2010).

2. Coupling mechanisms and normal-mode structure

In a two-ion chain inside a linear Paul trap, Coulomb repulsion converts single-particle motion into collective normal modes. For a $^{40}\mathrm{Ca}^+$–$^{40}\mathrm{CaH}^+$ crystal, the axial center-of-mass and breathing-mode frequencies obey

$$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$

with $\mu$ the mass ratio defined in the experiment and $\nu_1$ the single-ion axial secular frequency. Cooling one ion cools the shared $\bar n \ll 1$0 and $\bar n \ll 1$1 oscillators, and therefore the molecule as well (Rugango et al., 2014).

The same normal-mode logic extends to more complex crystals. In $\bar n \ll 1$2 pairs, the relevant axial modes are center-of-mass and stretch; in $\bar n \ll 1$3–$\bar n \ll 1$4, there are six secular modes; in strongly mismatched $\bar n \ll 1$5–$\bar n \ll 1$6 crystals, some radial modes are weakly coupled to the coolant because the cooling ion participates only weakly in them (Sriarunothai et al., 2017, Chen et al., 2016, King et al., 2021).

Hybrid platforms implement the same principle with different couplers. In cavity-mediated atom–membrane systems, the effective interaction takes the beam-splitter form

$\bar n \ll 1$7

so membrane phonons are coherently exchanged with an atomic center-of-mass mode that is continuously laser-cooled (Jöckel et al., 2014). In optoelectromechanical sympathetic cooling of an rf LC circuit, the optical cavity cools a nanomechanical oscillator, and the cooled mechanical mode then acts as a reservoir for the LC resonator through an electromechanical coupling $\bar n \ll 1$8, producing an indirect cooling path for the electrical degree of freedom (Malossi et al., 2020).

3. Representative realizations

The experimental record spans single molecules, clock ions, weakly coupled mixed crystals, and multi-ion chains. The table summarizes several benchmark demonstrations and near-ground-state regimes.

Platform	Coolant–target system	Reported motional result
Linear Paul trap	$\bar n \ll 1$9–$E_n=\hbar\omega(n+\tfrac12)$0	$E_n=\hbar\omega(n+\tfrac12)$1, $E_n=\hbar\omega(n+\tfrac12)$2 (Rugango et al., 2014)
Two-ion molecular crystal	$E_n=\hbar\omega(n+\tfrac12)$3–$E_n=\hbar\omega(n+\tfrac12)$4	$E_n=\hbar\omega(n+\tfrac12)$5, $E_n=\hbar\omega(n+\tfrac12)$6 (Wan et al., 2015)
Optical clock crystal	$E_n=\hbar\omega(n+\tfrac12)$7–$E_n=\hbar\omega(n+\tfrac12)$8	all six secular modes near the 3D ground state; secular time-dilation shift $E_n=\hbar\omega(n+\tfrac12)$9 at $\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$0 ms (Chen et al., 2016)
Weakly coupled mixed crystal	$\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$1–$\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$2	$\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$3, $\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$4, $\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$5 (King et al., 2021)
EIT sympathetic cooling	$\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$6–$\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$7 pair	$\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$8, $\bar n=\frac{1}{e^{\hbar\omega/k_BT}-1}.$9 (Lin et al., 2012)
RF-gradient sideband cooling	two $\bar n \ll 1$0 ions	target-ion COM mode $\bar n \ll 1$1 after sympathetic RF sideband cooling (Sriarunothai et al., 2017)

These results also expose the platform dependence of “ground-state cooling.” Some experiments explicitly report $\bar n \ll 1$2 from sideband asymmetry, while others establish translationally cold, crystallized, or near-ground-state conditions without full mode-by-mode quantum occupation readout. The contrast is especially clear when comparing phonon-resolved ion-chain work to translationally cold molecular-ion or recoil-ion spectroscopy platforms (Rugango et al., 2014, Wan et al., 2015, Zitzer et al., 2023).

4. Cooling methodologies

Resolved sideband cooling remains the canonical route in trapped ions. In the $\bar n \ll 1$3–$\bar n \ll 1$4 experiment, the coolant ion is driven on the $\bar n \ll 1$5 quadrupole transition at 729 nm, with 854 nm deshelving closing the cooling cycle. Alternating the first red sidebands of the center-of-mass and breathing modes for 6 ms cools both shared axial modes to near $\bar n \ll 1$6 (Rugango et al., 2014).

Raman sideband cooling is the dominant method in hyperfine qubit ions. For $\bar n \ll 1$7–$\bar n \ll 1$8, robust cooling is obtained by combining pulsed Raman sideband cooling with continuous quench cooling. The scheme uses higher-order red sidebands to remove high-$\bar n \ll 1$9 population and then first-order sidebands to finish near the ground state, and it remains effective even outside the Lamb–Dicke regime, where couplings develop $^{40}\mathrm{Ca}^+$0-dependent zeros (Wan et al., 2015). The $^{40}\mathrm{Ca}^+$1–$^{40}\mathrm{Ca}^+$2 clock implementation similarly exploits Raman sideband cooling on all six secular modes, but its analysis emphasizes non-thermal Fock-state distributions and conservative motional-energy bounds rather than simple thermal fits (Chen et al., 2016).

Electromagnetically induced transparency cooling provides a broadband alternative. In $^{40}\mathrm{Ca}^+$3–$^{40}\mathrm{Ca}^+$4 chains, EIT cooling applied only to $^{40}\mathrm{Ca}^+$5 cools several axial modes simultaneously and reduces the required cooling time by approximately an order of magnitude relative to conventional Raman sideband cooling, with a significant reduction in required laser intensity (Lin et al., 2012).

Radio-frequency sideband cooling in a static magnetic-field gradient replaces optical sideband coupling with an effective Lamb–Dicke parameter generated by the position dependence of the Zeeman shift. In $^{40}\mathrm{Ca}^+$6, a static gradient of $^{40}\mathrm{Ca}^+$7 yields $^{40}\mathrm{Ca}^+$8, sufficient for RF red-sideband cooling of both single-ion motion and the center-of-mass mode of a two-ion crystal, including the first realisation of sympathetic RF sideband cooling of two individually addressable identical ions (Sriarunothai et al., 2017).

When direct sympathetic cooling is weak because the coolant barely participates in a target mode, phonons can be rerouted coherently. The $^{40}\mathrm{Ca}^+$9–$^{40}\mathrm{CaH}^+$0 work introduces an algorithmic protocol in which radial phonons are mapped onto the highly charged ion’s internal state, transferred into a strongly cooled axial mode, and then dissipated through $^{40}\mathrm{CaH}^+$1. This is a quantum-logic realization of sympathetic cooling by phonon transfer rather than by direct sideband coupling to the problematic mode (King et al., 2021).

5. Thermometry, heating, and non-ideal behavior

The standard trapped-ion thermometry observable is the red/blue sideband asymmetry. In the Lamb–Dicke regime and for a thermal distribution,

$^{40}\mathrm{CaH}^+$2

This relation underlies the extraction of $^{40}\mathrm{CaH}^+$3 in molecular-ion and many atomic-ion experiments (Rugango et al., 2014). Yet sideband thermometry can become misleading once the distribution ceases to be thermal. In the $^{40}\mathrm{CaH}^+$4–$^{40}\mathrm{CaH}^+$5 clock, the post-cooling distribution contains a tiny high-$^{40}\mathrm{CaH}^+$6 plateau carrying disproportionate kinetic energy, motivating a double-thermal analysis and conservative $^{40}\mathrm{CaH}^+$7 upper bounds on $^{40}\mathrm{CaH}^+$8 (Chen et al., 2016).

Motional heating is the dominant practical limit after cooling. In the $^{40}\mathrm{CaH}^+$9–$$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$0 system, the measured axial heating rates are $$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$1 phonons/ms for a single $$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$2 ion and $$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$3 phonons/ms for the two-ion center-of-mass mode (Rugango et al., 2014). In the $$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$4–$$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$5 clock, the six-mode heating rates range from $$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$6 quanta/s for z-STR to $$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$7 quanta/s for x-COM (Chen et al., 2016). In the $$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$8–$$\nu_{\pm}^2=\left[(1+\mu)\pm\sqrt{1-\mu+\mu^2}\right]\nu_1^2,$$9 HCI platform, the weakly coupled radial mode at $\mu$0 MHz corresponds to an electric-field noise spectral density $\mu$1, reported there as the lowest observed electric-field noise spectral density for a radiofrequency ion trap (King et al., 2021).

The literature also identifies several non-ideal sympathetic-cooling regimes. In atom–ion collisions inside an rf Paul trap, collisions can either cool or heat because the oscillating trap fields drive a non-equilibrium dynamics; direct single-shot measurements show heating events up to about 350 K in the radial regime, even though the bath is ultracold (Meir et al., 2018). For neutral polar molecules, trap architecture can dominate the outcome: in a static electric trap inelastic losses are too great for LiH + Li cooling to be feasible, an ac electric trap adds trajectory-instability loss, while in a microwave trap there are no such losses and sympathetic cooling should be possible (Tokunaga et al., 2010).

A recurring misconception is that a Coulomb crystal, a translationally cold ensemble, or a narrow fluorescence image automatically implies quantum ground-state motion. The molecular-ion preparation of $\mu$2 reports translational energies of $\mu$3 mK for Ca$\mu$4 and $\mu$5 mK for $\mu$6, and the Th$\mu$7 recoil-ion spectroscopy platform infers an upper-bound temperature of about 140 mK from Doppler broadening; both are valuable sympathetic-cooling regimes, but neither is equivalent to the $\mu$8 conditions explicitly demonstrated in sideband-resolved experiments (Tong et al., 2010, Zitzer et al., 2023).

6. Scientific uses, platform extensions, and outlook

In precision spectroscopy, sympathetic ground-state cooling suppresses Doppler broadening, reduces time-dilation shifts, and enables quantum-logic readout. This is explicit in the $\mu$9–$\nu_1$0 optical clock, where near-3D ground-state cooling yields a secular-motion time-dilation shift of $\nu_1$1 for a typical $\nu_1$2 ms probe, a 50-fold reduction in the shift uncertainty relative to earlier $\nu_1$3 clocks (Chen et al., 2016). In molecular-ion work, the same level of motional control is identified as the starting point for quantum-logic-assisted internal-state preparation, detection, and spectroscopy (Wan et al., 2015).

For molecular physics, sympathetic cooling underpins two distinct agendas. One is high-precision internal-state spectroscopy of trapped molecules and molecular ions, including searches for variation of the electron-to-proton mass ratio and other tests of fundamental physics (Rugango et al., 2014). The other is state-preserving preparation of translationally cold, rovibrationally selected ions: threshold-photoionized $\nu_1$4 with $\nu_1$5 selectivity in the ground rovibrational level and state lifetimes on the order of 15 minutes was presented as a route toward cold quantum-controlled ion–molecule collision studies, frequency-metrology experiments, and molecular-ion qubits (Tong et al., 2010). The sympathetically cooled Th$\nu_1$6 recoil-ion platform extends this logic to nuclear spectroscopy, where the system is being developed for hyperfine spectroscopy of electronic transitions in $\nu_1$7 relevant to the nuclear ground and isomeric states (Zitzer et al., 2023).

The scope of sympathetic ground-state cooling now extends well beyond trapped-ion crystals. In an atom–membrane hybrid system, ultracold $\nu_1$8 sympathetically cooled a Si$\nu_1$9N$\bar n \ll 1$00 membrane from room temperature to $\bar n \ll 1$01 mK, and the analysis identified a cooperativity criterion $\bar n \ll 1$02 for ground-state operation, with realistic improvements giving $\bar n \ll 1$03 (Jöckel et al., 2014). For levitated fused-silica nanospheres optically coupled to cold atoms, the predicted steady-state occupations are $\bar n \ll 1$04 for a 300 nm sphere and $\bar n \ll 1$05 for a 100 nm sphere, explicitly placing sympathetic ground-state cooling in a room-temperature optomechanical setting (Ranjit et al., 2014). In optoelectromechanics, a macroscopic rf LC resonator can be sympathetically cooled to its ground state when the optomechanical cooperativity is large, the electromechanical cooperativity is larger, and comparable optomechanical and electromechanical coupling rates are realized (Malossi et al., 2020).

A plausible synthesis of these developments is that sympathetic ground-state cooling is best viewed not as a single technique but as a design principle: engineer a strongly coupled target–refrigerator pair, ensure that the refrigerator thermalizes rapidly to a low-entropy bath, and suppress mechanisms that re-inject energy faster than shared modes can be emptied. In trapped ions this usually means large coolant participation, resolved or engineered sidebands, and low electric-field noise; in neutral and hybrid systems it means favorable elastic-to-inelastic ratios, collisionally stable trap geometries, or sufficiently large hybrid cooperativity. The breadth of the current literature suggests that the concept is now a unifying control primitive for systems ranging from molecular ions and optical clocks to membranes, levitated nanoparticles, and macroscopic electrical resonators (Lin et al., 2012, Sriarunothai et al., 2017, Christoph et al., 2018, King et al., 2021).