Quantum Droplets: Self-Bound Quantum Liquids

• Quantum droplets are self-bound many-body quantum liquids where attractive mean-field forces are balanced by repulsive quantum fluctuations, leading to liquid-like states.
• Their stabilization mechanisms rely on Lee–Huang–Yang corrections or alternative fluctuation-induced terms across various geometries and dimensions.
• Experimental realizations in Bose mixtures and dipolar condensates reveal flat-top density profiles, finite-size behavior, and phase-coherent dynamics.

Quantum droplets are self-bound, ultradilute quantum liquids in which a Bose gas remains localized in free space because an attractive mean-field contribution is balanced by beyond-mean-field quantum fluctuations. In the established ultracold-atom realizations, this balance is most often provided by a Lee–Huang–Yang (LHY) correction that stabilizes either a binary Bose mixture or a dipolar condensate at a finite equilibrium density, producing liquid-like states with flat-top density profiles, weak compressibility, and surface tension rather than trap-determined gaseous clouds (Luo et al., 2020, Cabrera et al., 2017, Ferrier-Barbut et al., 2016). Subsequent work has extended the concept to reduced dimensions, rotating states with vortices, capillary breakup into multiple droplets, cavity-mediated liquids, curved geometries, polaritonic fluids, and quadrupolar condensates.

1. Definition, thermodynamic criteria, and conceptual scope

A quantum droplet is a self-bound many-body state with a finite equilibrium volume in the absence of an external trap. In thermodynamic language, a self-bound droplet satisfies three conditions: zero pressure at equilibrium, positive compressibility, and a negative effective chemical potential at the equilibrium volume,

$$P_0=-\left(\frac{\partial E_0}{\partial V}\right)_N=0,\qquad \left(\frac{\partial^2 E_0}{\partial V^2}\right)_N\Big|_{V_0}>0,\qquad \left(\frac{\partial E_0}{\partial N}\right)_{V_0}<0.$$

These criteria distinguish a droplet from a trapped condensate, which expands when confinement is removed, and from a collapsing mean-field state, which has no finite-density minimum (Mixa et al., 2024).

The term “liquid-like” is used in a precise sense. In the droplet regime, the density is set predominantly by the internal balance of interactions and quantum fluctuations, not by the trap. As atom number increases, the central density tends to saturate and the droplet grows mainly by increasing its size, which is the characteristic finite-size behavior of a compressible quantum liquid rather than a gas (Cabrera et al., 2017, Ferrier-Barbut et al., 2016).

A recurrent misconception is to identify quantum droplets with bright solitons. In one dimension, the crossover can be smooth, and small droplets may look Gaussian and collide quasi-elastically, but the defining stabilization mechanism is different. Bright solitons are mean-field objects stabilized by kinetic dispersion against attraction, whereas quantum droplets are beyond-mean-field objects whose existence depends on fluctuation-induced terms in the equation of state (Astrakharchik et al., 2018).

2. Stabilization mechanisms and equations of state

The canonical Bose–Bose mechanism is the Petrov scenario: intraspecies interactions are repulsive, interspecies interactions are attractive, and the net mean-field interaction is tuned to be weakly attractive. Mean-field theory alone would then predict collapse, because the energy density scales as $n^2$. Quantum fluctuations generate an LHY contribution that scales as $n^{5/2}$ in three dimensions and becomes repulsive at higher density, halting the collapse and creating a minimum of the energy per particle at finite density (Cabrera et al., 2017, Mixa et al., 2024).

Dipolar droplets realize the same structural idea with different microscopic ingredients. There the competition is between short-range contact repulsion and anisotropic dipole–dipole interactions, which can produce an effectively attractive channel for suitable geometry. In the dysprosium experiments and theory, the mean-field contribution becomes weakly attractive for elongated clouds, while the LHY correction remains repulsive and stabilizes a self-bound droplet with a finite saturation density (Ferrier-Barbut et al., 2016, Ferrier-Barbut et al., 2016).

Dimensional reduction changes the form of the stabilizing term. In two dimensions, the effective beyond-mean-field contribution is encoded by a logarithmic nonlinearity, so that the energy density contains a term proportional to $n^2\ln n$, and the extended Gross–Pitaevskii equation contains $|\psi|^2\ln|\psi|^2$ rather than a simple power law. In one dimension, the beyond-mean-field contribution changes character again; for the symmetric binary mixture studied by Astrakharchik and Malomed, the effective equation contains cubic repulsion together with quadratic attraction, and the exact stationary droplet solution interpolates between small Gaussian droplets and large flat-top “puddles” (Tengstrand et al., 2019, Pelayo et al., 2024, Astrakharchik et al., 2018).

The existence of alternative stabilization mechanisms does not alter the defining logic. The cavity-induced proposal is explicit on this point: even a mean-field-stable, purely repulsive condensate can become self-bound when cavity-induced roton quantum fluctuations contribute a negative pressure that competes with the positive contact-interaction pressure. In that case, the droplet belongs to a different finite-size class, but the self-bound state is still produced by a competition between a mean-field term and a fluctuation term with a more favorable volume scaling (Mixa et al., 2024).

3. Dimensionality, effective theories, and beyond-LHY descriptions

Quantum-droplet theory is strongly dimension-dependent. In the three-dimensional contact-mixture problem, the effective description is the extended Gross–Pitaevskii equation with a local LHY term. In strictly two-dimensional or quasi-two-dimensional settings, the effective energy functional becomes logarithmic, and the crossover from 3D to 2D must be treated explicitly. In the $^{39}\mathrm{K}$ crossover analysis, the pure 2D model and the quasi-2D model agree only in a specific region of parameter space; one of the main findings is that droplets become substantially extended when the averaged mean-field interaction changes from negative to positive, with binding energies approximately inversely proportional to the square of the size (Pelayo et al., 2024).

One-dimensional theories are more delicate. The amended 1D Gross–Pitaevskii equation

$$i\psi_t+\frac{1}{2}\psi_{xx}-|\psi|^2\psi+|\psi|\psi=0$$

supports an exact droplet solution and reveals two physically distinct regimes: small droplets of approximately Gaussian shape and large flat-top puddles. In the quasi-1D dipolar problem, the effective equation is nonlocal, with a reduced dipolar kernel and a local LHY correction proportional to $|\psi|^3$; this geometry supports quasi-1D droplets, multiple roton instabilities, and a droplet–soliton crossover that depends sensitively on polarization angle and confinement (Astrakharchik et al., 2018, Edmonds et al., 2020).

Standard extended-GP theory is not the endpoint of the subject. A density-functional formulation has been developed in which the correlation energy is generated self-consistently from an effective action. In that approach the correlation energy is real, higher-order corrections to the ground-state energy and quantum depletion can be computed, and the familiar imaginary-part problem of the Bogoliubov LHY energy in the droplet regime is avoided by renormalizing the effective couplings self-consistently (Zhang et al., 2023).

Ab-initio many-body calculations show where beyond-LHY physics becomes quantitatively important. In one-dimensional homonuclear and heteronuclear mixtures, ML-MCTDHX simulations demonstrate correlation holes at the mean-field balance point and in flat-top droplets, reduced expansion velocities and breathing frequencies relative to the modified Gross–Pitaevskii theory, and explicit signatures of intercomponent entanglement and two-body anti-correlations at the same position together with longer-range correlations (Mistakidis et al., 2021).

4. Experimental realizations in ultracold gases

The first dipolar observations in $^{164}\mathrm{Dy}$ established stable droplets containing $\sim 800$ atoms in a regime where mean-field theory predicts collapse, and they also showed interference of several droplets, indicating phase coherence of the droplet state (Ferrier-Barbut et al., 2016). A complementary dysprosium study emphasized that these are liquid-like droplets of ultracold magnetic atoms stabilized by quantum fluctuations, with densities in the droplet regime around $n^2$0, markedly higher than the gas-like BEC densities, and with inter-droplet repulsion governed by the dipolar interaction (Ferrier-Barbut et al., 2016).

Contact-only droplets were then observed in a $n^2$1 Bose mixture. These droplets are more than eight orders of magnitude more dilute than liquid helium, are self-bound in the plane where confinement is removed, and exhibit a minimum atom number below which quantum pressure drives a liquid-to-gas transition. The same study showed directly, by comparison with a single-component condensate, that quantum many-body effects stabilize the mixture droplet against collapse (Cabrera et al., 2017).

Heteronuclear droplets have enlarged the phenomenology. In a $n^2$2–$n^2$3 mixture released in an optical waveguide, a sudden change of the interspecies interaction from the non-interacting to the strongly attractive regime produces an initially single droplet in an excited compression–elongation mode. That filament stretches to a critical length and then breaks into two or more smaller droplets, a process consistent with a capillary instability governed by surface tension (Cavicchioli et al., 2024).

Experimental observables are correspondingly diverse. Density profiles reveal flat-top bulk regions and anisotropic deformation; time-of-flight or in-situ expansion distinguishes self-bound droplets from expanding condensates; lifetime scaling with scattering length tests the role of LHY stabilization; interference fringes establish phase coherence; and inter-droplet spacing, motion, or fragmentation probe effective interactions and surface dynamics (Ferrier-Barbut et al., 2016, Cabrera et al., 2017, Cavicchioli et al., 2024).

5. Collective excitations, vortices, fragmentation, and nonlinear dynamics

Rotation introduces quantized vorticity into the droplet problem. In a two-dimensional binary droplet described by the logarithmic 2D extended GP equation, increasing the rotation frequency produces successive entries of singly quantized vortices. For $n^2$4 and $n^2$5, the angular-momentum curve exhibits steps at $n^2$6, $n^2$7, and $n^2$8, corresponding to one-, two-, and three-vortex states. These small vortex clusters form finite-size “precursors of an Abrikosov lattice” and can remain self-bound for relevant times after trap removal (Tengstrand et al., 2019).

One-dimensional dynamics shows a strong size dependence. Small droplets collide quasi-elastically and behave in a soliton-like way, whereas large colliding puddles may merge or fragment depending on relative velocity. The breakup threshold is well organized by a Weber-number argument: for large droplets, fragmentation occurs when the kinetic energy associated with a perturbation becomes comparable to the surface energy (Astrakharchik et al., 2018).

Dipolar quasi-1D droplets enrich this picture further. Their Bogoliubov spectrum can develop roton minima, and with LHY corrections there can be two roton-instability regions as a function of dipolar strength. Interaction quenches trigger modulational instability, generating multiple droplets, bright solitons, and radiation, while collisions exhibit population transfer, fission, and phase-sensitive outcomes consistent with Josephson-like behavior during the interaction time (Edmonds et al., 2020).

The space of excited droplet states is also broader than the monopole sector. In quasi-1D asymmetric mixtures, multipole droplets exist in which one component changes sign and carries dipole or tripole structure while the other remains nodeless. These states have no counterpart in the reduced single-component model for symmetric mixtures, and large parts of their existence domain are linearly and dynamically stable (Kartashov et al., 2024).

Beyond-LHY many-body dynamics modifies even basic collective modes. In one-dimensional droplets, quench dynamics toward stronger or weaker attractions shows that fully correlated calculations yield reduced breathing frequencies and slower expansion than the modified GP theory, while the two-body correlation function develops anti-correlations at the same position and longer-range correlations across the droplet (Mistakidis et al., 2021).

6. Emerging platforms, geometries, and generalizations

Recent theory has expanded quantum-droplet physics beyond flat-space atomic mixtures. In an optical cavity with transverse pumping, a mean-field-stable Bose gas can form a self-bound droplet because cavity-induced long-range interactions soften a roton mode, and the corresponding quantum fluctuation energy contributes a negative pressure. These cavity-induced droplets are finite-size objects whose equilibrium volume scales as $n^2$9, and they were explicitly identified as belonging to a distinct “D3” class of droplet mechanisms (Mixa et al., 2024).

Curved backgrounds introduce a different extension. In the curved-space formulation of a Bose–Bose mixture, the one-loop effective action acquires curvature-dependent terms involving $n^{5/2}$0 and mixed $n^{5/2}$1 structures. In the two-dimensional case treated in detail, positive curvature deepens the finite-density minimum and increases the peak density, whereas sufficiently negative curvature can remove the non-zero minimum entirely through a first-order quantum phase transition. Numerical examples on a weakly curved spherical cap and a negatively curved surface show peak densities shifting from $n^{5/2}$2 in flat space to $n^{5/2}$3 for $n^{5/2}$4 and $n^{5/2}$5 for $n^{5/2}$6 (Flachi et al., 18 Jul 2025).

Solid-state and higher-multipole realizations extend the same self-bound logic to new quasiparticles and interactions. A spin mixture of exciton-polaritons near a biexciton Feshbach resonance is predicted to support self-bound “quantum droplets of light” in a detuning window of about $n^{5/2}$7, with saturation densities $n^{5/2}$8 and radii $n^{5/2}$9 (Caldara et al., 25 Jul 2025). In quasi-2D condensates with quadrupole–quadrupole interactions, Thomas–Fermi theory and numerics predict flat-top droplets, linear scaling of effective area with particle number, saturation of peak density and chemical potential at large norm, and anisotropic elliptical vortex droplets whose aspect ratios are tunable by norm and quadrupolar interaction strength (Xia et al., 15 Oct 2025).

These developments do not replace the standard mixture and dipolar paradigms; rather, they show that “quantum droplet” now denotes a broader class of self-bound, fluctuation-stabilized liquids whose detailed equation of state, symmetry, and collective dynamics depend on dimensionality, geometry, and the microscopic origin of the attractive channel.