Critical Casimir Force

• Critical Casimir Force (CCF) is the effective interaction generated by confining long-range fluctuations near a continuous phase transition.
• Its sign, range, and magnitude depend on geometry, boundary conditions, and universality classes, with examples in binary mixtures, superfluid films, and membranes.
• Both experimental and theoretical studies use scaling laws and optical-tweezer methods to measure CCF, enabling applications in colloidal assembly and micro-nano device engineering.

Critical Casimir force (CCF) is the effective interaction generated when boundaries, colloids, or inclusions confine the long-ranged fluctuations of an order parameter in a medium near a continuous phase transition. In binary liquids near demixing, the order parameter is the local concentration deviation; in superfluid films it is the complex superfluid field; in membranes at miscibility criticality it is the local composition contrast. The effect is the thermal analogue of the quantum electromagnetic Casimir effect, but its sign, range, and magnitude are controlled by universality class, geometry, and boundary conditions imposed by surface adsorption preferences, while its range is set by the bulk correlation length $\xi$ that diverges at criticality (Callegari et al., 2020, Gambassi et al., 2023).

1. Physical origin and universality structure

The basic mechanism is confinement-induced modification of critical fluctuations. Near the critical temperature $T_c$, the bulk correlation length obeys

$$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$

with $\nu \approx 0.63$ for the three-dimensional Ising universality class relevant to binary liquid mixtures, and $\nu \approx 0.67$ for the three-dimensional XY universality class relevant to superfluid ${}^4\mathrm{He}$ films (Gambassi et al., 2023). In water–2,6-lutidine, values $\xi_0^+ \approx 0.2\,\mathrm{nm}$ or $\xi_0 \approx 0.22\,\mathrm{nm}$ are used in direct colloidal and metallic-body experiments, whereas larger amplitudes such as $\xi_0 \approx 1.4\,\mathrm{nm}$ occur in micellar systems like $\mathrm{C}_{12}\mathrm{E}_5$–water, making CCF substantially longer-ranged at the same reduced temperature (Magazzù et al., 2018, Schmidt et al., 2022, Callegari et al., 2020).

Universality fixes the scaling functions once the bulk and surface universality classes are specified. For binary mixtures, the order parameter is proportional to the concentration difference; in field-theoretic notation the bulk behavior is captured by a Landau–Ginzburg–Wilson functional with bulk couplings and surface fields $T_c$0, surface enhancements $T_c$1, and the ordinary, extraordinary, and special surface universality classes emerging as RG fixed points (Gambassi et al., 2023). Strong preferential adsorption yields the symmetry-breaking $T_c$2 and $T_c$3 boundary conditions, while Dirichlet-like behavior corresponds to the ordinary class and is often denoted $T_c$4 in superfluid-film contexts (Gambassi et al., 2023).

The sign of the force is set chiefly by boundary conditions. Symmetric boundary conditions such as $T_c$5, $T_c$6, or, for superfluid films, $T_c$7 generally yield attraction, whereas antisymmetric $T_c$8 conditions yield repulsion (Callegari et al., 2020, Gambassi et al., 2023). In near-critical solvents this mapping is operational: hydrophilic surfaces prefer the water-rich phase, hydrophobic surfaces prefer the lutidine-rich phase in water–lutidine, and appropriate surface treatments therefore switch the force from attractive to repulsive (Callegari et al., 2020).

2. Scaling forms, geometries, and exact results

In film geometry, the singular excess free energy per area and the corresponding force per area admit universal finite-size scaling forms. One standard representation is

$T_c$9

which in $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$0 becomes the familiar $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$1 (Gambassi et al., 2023). In the notation of exact finite-size scaling theory, the dimensionless excess free energy $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$2 and dimensionless Casimir pressure $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$3 satisfy

$$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$4

with scaling variable $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$5 and critical Casimir amplitude $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$6 at $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$7 (Dantchev et al., 2022).

For curved objects, the Derjaguin or proximity-force approximation maps the problem to film geometry. For a sphere of radius $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$8 at surface-to-surface distance $$\xi(t)=\xi_0 |t|^{-\nu}, \qquad t=\frac{T-T_c}{T_c},$$9 from a plate,

$\nu \approx 0.63$0

while for two spheres one replaces $\nu \approx 0.63$1 by the effective radius $\nu \approx 0.63$2 (Gambassi et al., 2023, Callegari et al., 2020). In planar metallic platelet geometry, by contrast, the directly relevant form is

$\nu \approx 0.63$3

which was used to quantify the competition between critical Casimir and Casimir–Lifshitz forces for gold flakes above gold-coated substrates (Schmidt et al., 2022).

Geometry	Universal form	Remarks
Film	$\nu \approx 0.63$4; $\nu \approx 0.63$5	Sign fixed by BC
Sphere–plate	$\nu \approx 0.63$6	Derjaguin regime $\nu \approx 0.63$7
Sphere–sphere	Same form with $\nu \approx 0.63$8	Pair picture breaks down near many-body regime

Exact results are especially developed in two dimensions. For strips in the 2D Ising universality class, the lateral force generated by chemically inhomogeneous walls has the exact scaling form

$\nu \approx 0.63$9

and the leading term is independent of the magnitude of $\nu \approx 0.67$0 provided $\nu \approx 0.67$1 is finite (Dantchev et al., 2022). Conformal field theory further shows that alternating boundary conditions can generate stable equilibrium positions, surface RG flows between distinct boundary fixed points, and lateral forces with a universal simple cosine form at large separations (Dubail et al., 2016).

3. Experimental realizations and measurement methodologies

Experimental access to CCF has progressed from wetting films to direct force spectroscopy on colloids and microstructures. Indirect evidence came from film thinning in $\nu \approx 0.67$2 and thickening or thinning in binary mixtures depending on whether the force was repulsive $\nu \approx 0.67$3 or attractive $\nu \approx 0.67$4, while direct measurements were established in sphere–plate and sphere–sphere geometries in near-critical solvents (Gambassi et al., 2023).

Optical trapping became central because CCF act over tens to hundreds of nanometers, a range that is simultaneously short-ranged and experimentally delicate. In optical-tweezer experiments, a trapped particle experiences

$\nu \approx 0.67$5

with trap stiffness calibrated by equipartition or spectral methods. The total interaction is then reconstructed from the equilibrium distribution

$\nu \approx 0.67$6

and noncritical contributions such as electrostatics, gravity or buoyancy, van der Waals forces, and near-wall hydrodynamics are subtracted to isolate $\nu \approx 0.67$7 (Callegari et al., 2020). In sphere–plate experiments, Total Internal Reflection Microscopy yields nanometer-resolution heights via the exponentially decaying evanescent intensity, whereas digital video microscopy is used for bulk multi-particle tracking (Callegari et al., 2020).

A canonical direct-measurement system is water–2,6-lutidine at critical composition $\nu \approx 0.67$8, with lower critical temperature reported as $\nu \approx 0.67$9 in TIRM sphere–plate studies. There, hydrophilic spheres near hydrophilic substrates show increasing attraction on approaching ${}^4\mathrm{He}$0, whereas hydrophobic spheres near hydrophilic substrates show increasing repulsion; separations ${}^4\mathrm{He}$1–${}^4\mathrm{He}$2 are probed with femtonewton force resolution (Callegari et al., 2020). In ${}^4\mathrm{He}$3–water, micron-scale spheres held in two traps at ${}^4\mathrm{He}$4 separations display strongly enhanced coupling because ${}^4\mathrm{He}$5 reaches about ${}^4\mathrm{He}$6 at ${}^4\mathrm{He}$7, compared with about ${}^4\mathrm{He}$8 in water–lutidine under similar conditions (Callegari et al., 2020).

CCF also couple directly to quantum-electrodynamic fluctuation forces. In experiments on hydrophilic gold microflakes above gold-coated substrates immersed in water–2,6-lutidine, attractive Casimir–Lifshitz forces dominate far from ${}^4\mathrm{He}$9 and cause stiction, but antisymmetric $\xi_0^+ \approx 0.2\,\mathrm{nm}$0 boundary conditions generate repulsive critical Casimir forces near $\xi_0^+ \approx 0.2\,\mathrm{nm}$1 that lift the flake from $\xi_0^+ \approx 0.2\,\mathrm{nm}$2 to $\xi_0^+ \approx 0.2\,\mathrm{nm}$3 and counteract stiction; the experimentally inferred height has $\xi_0^+ \approx 0.2\,\mathrm{nm}$4 uncertainty (Schmidt et al., 2022). Patterned substrates extend this control further: a bull’s-eye pattern with a ring of opposite surface preference supports sedimentation to the ring, ring levitation, and point levitation above the center, with point-levitation regions shrinking as the system is driven away from criticality even though the trapping force becomes stronger (Nowakowski et al., 2024).

4. Many-body effects, dynamics, and force fluctuations

Although pairwise Derjaguin pictures are often effective, CCF are intrinsically nonadditive. Holographic-tweezer experiments in water–lutidine provided the first clear experimental evidence of many-body critical Casimir effects: the effective two-body attraction between two hydrophilic spheres is reduced when a third nearby particle is added, and the deviation from pairwise additivity grows as $\xi_0^+ \approx 0.2\,\mathrm{nm}$5 (Callegari et al., 2020). The broader colloidal literature likewise emphasizes that effective one-component descriptions are reliable mainly for dilute suspensions or modest $\xi_0^+ \approx 0.2\,\mathrm{nm}$6, whereas dense suspensions and confined geometries require explicit treatment of the ternary solvent–colloid mixture (Maciolek et al., 2017).

Dynamical manifestations are now well established. In blinking-tweezer experiments on two silica colloids in near-critical water–lutidine, the attractive CCF modifies drift velocities, relative diffusion, and first-passage statistics. The pair potential is modeled as

$\xi_0^+ \approx 0.2\,\mathrm{nm}$7

with $\xi_0^+ \approx 0.2\,\mathrm{nm}$8 extracted from equilibrium separation histograms, and the resulting Langevin dynamics reproduces the temperature dependence of first-passage times: when trajectories start within the attractive basin, the mean escape time grows as $\xi_0^+ \approx 0.2\,\mathrm{nm}$9 increases from roughly $\xi_0 \approx 0.22\,\mathrm{nm}$0 to $\xi_0 \approx 0.22\,\mathrm{nm}$1 (Magazzù et al., 2018). Related optical-tweezer experiments in micellar solvents showed CCF-induced synchronization of colloidal motions and temperature-dependent work and heat statistics (Gambassi et al., 2023).

Theoretical descriptions of nonequilibrium CCF use dynamic universality classes rather than static scaling alone. Model A and Model B stochastic dynamics describe nonconserved and conserved order parameters, respectively, while Model H couples a conserved order parameter to hydrodynamics in binary fluids (Gambassi et al., 2023). Fluid-particle-dynamics simulations of moving colloids in near-critical binary mixtures showed that advection and critical slowing down produce retarded, weakened, and anisotropic effective CCF whenever the Péclet number becomes appreciable; even the drag on a single colloid is enhanced by the critical adsorption cloud (Furukawa et al., 2013).

Because the force originates in fluctuations, it fluctuates itself. Within Gaussian Model B, the instantaneous force can be defined rigorously through the dynamic stress tensor, and its static variance is dominated by ultraviolet modes and therefore depends on the microscopic cutoff, whereas the dynamic force–force correlation at nonzero time is cutoff-independent and decays algebraically in time with an exponent determined by the dynamic universality class rather than by boundary conditions within the Gaussian approximation (Gross et al., 2020). For a fluctuating film thickness, this induces colored, effectively non-Markovian noise in the boundary dynamics and generates an algebraic short-time contribution to the position variance before long-time saturation (Gross et al., 2020).

5. Patterning, disorder, ensemble dependence, and notable caveats

Spatially heterogeneous boundaries enrich the phenomenology beyond the simple $\xi_0 \approx 0.22\,\mathrm{nm}$2 versus $\xi_0 \approx 0.22\,\mathrm{nm}$3 dichotomy. Chemically stepped and striped substrates generate lateral as well as normal CCF, and within the Derjaguin approximation the force can even change sign as a function of sphere–substrate distance, enabling stable levitation with temperature sensitivities exceeding $\xi_0 \approx 0.22\,\mathrm{nm}$4 in representative water–lutidine parameters (Tröndle et al., 2010). In two dimensions, alternating boundary conditions generate exact lateral-force scaling functions and surface RG flows between free and symmetry-breaking fixed points (Dubail et al., 2016).

Quenched surface disorder provides a second layer of structure. In a three-dimensional Ising film with one homogeneous $\xi_0 \approx 0.22\,\mathrm{nm}$5 wall and an opposing wall with random local adsorption preference, tuning the disorder parameter $\xi_0 \approx 0.22\,\mathrm{nm}$6 drives a crossover from repulsive $\xi_0 \approx 0.22\,\mathrm{nm}$7 behavior to attractive $\xi_0 \approx 0.22\,\mathrm{nm}$8 behavior, while $\xi_0 \approx 0.22\,\mathrm{nm}$9 realizes an effective Dirichlet or ordinary boundary condition that is normally inaccessible in classical fluids (Toldin, 2013). For films in the ordinary surface universality class with zero-mean Gaussian random surface fields on both walls, the disorder scaling field

$\xi_0 \approx 1.4\,\mathrm{nm}$0

is RG-irrelevant asymptotically in $\xi_0 \approx 1.4\,\mathrm{nm}$1, but for thin films with $\xi_0 \approx 1.4\,\mathrm{nm}$2 it significantly deepens the attractive minimum and shifts it to lower temperatures while leaving the force attractive (Maciolek et al., 2015).

The force can also depend qualitatively on the thermodynamic ensemble. For Dirichlet boundary conditions in mean-field theory, the grand-canonical ensemble with fixed bulk field yields an attractive CCF, whereas the canonical ensemble with fixed total order parameter typically yields a repulsive CCF; Monte Carlo simulations of the three-dimensional Ising model confirm grand-canonical attraction and canonical repulsion for sufficiently large fixed magnetization (Rohwer et al., 2018). This is not a small correction but a sign reversal driven by the global constraint.

A further caveat comes from models with long-range couplings. In the finite-size Nagle–Kardar model with periodic boundary conditions, the exact Casimir amplitudes are $\xi_0 \approx 1.4\,\mathrm{nm}$3 on the critical line and $\xi_0 \approx 1.4\,\mathrm{nm}$4 at the tricritical point, and the force is repulsive near both singular regimes before becoming attractive away from them (Dantchev et al., 2024). This model-specific result violates the widely accepted “boundary condition rule” that equivalent boundary conditions should imply attraction, and it demonstrates that the sign structure of CCF is not universally reducible to boundary symmetry alone (Dantchev et al., 2024).

6. Applications, related domains, and open problems

The practical importance of CCF lies in their reversibility and fine temperature control. In colloidal suspensions they drive thermally reversible aggregation, gas–liquid–solid transitions, face-centered-cubic ordering, and directional assembly with Janus or patchy particles; in patterned environments they generate lateral trapping, torques, and anisotropic interactions useful for reconfigurable architectures (Maciolek et al., 2017, Gambassi et al., 2023). In micro- and nanomechanical systems, repulsive CCF provide a liquid-phase anti-stiction mechanism that counteracts Casimir–Lifshitz attraction without requiring specialized optical materials (Schmidt et al., 2022). Patterned bull’s-eye substrates further suggest size-selective colloid sorting and “critical Casimir thermometry,” because the levitation window depends sharply on $\xi_0 \approx 1.4\,\mathrm{nm}$5 (Nowakowski et al., 2024).

The concept extends beyond soft colloids. In two-dimensional cellular membranes near miscibility criticality, conformal field theory predicts composition-mediated forces between membrane inclusions: fixed–fixed boundary conditions yield

$\xi_0 \approx 1.4\,\mathrm{nm}$6

at criticality, whereas interactions involving free boundary conditions decay as $\xi_0 \approx 1.4\,\mathrm{nm}$7 (Machta et al., 2012). In magnetic systems, a thermodynamic protocol based on integrating the order parameter along isotherms has been proposed to extract CCF directly from thin-film magnetization data, with the additional observation that a bulk field opposing boundary ordering can enhance the force by more than an order of magnitude relative to zero-field conditions (Cardozo et al., 2014). For periodic boundaries, the classical CCF can even be reconstructed from the internal-energy equation of state of a symmetry-related $\xi_0 \approx 1.4\,\mathrm{nm}$8 quantum critical system, providing a route through quantum Monte Carlo and cold-atom or solid-state experiments to scaling functions otherwise difficult to measure (Rancon et al., 2016).

Several issues remain open. Precise temperature control at the millikelvin level, suppression or calibration of optical heating, and careful subtraction of electrostatic, hydrodynamic, and dispersion backgrounds remain indispensable near $\xi_0 \approx 1.4\,\mathrm{nm}$9 (Callegari et al., 2020, Schmidt et al., 2022). Finite adsorption strength, often encoded through a surface ordering field $\mathrm{C}_{12}\mathrm{E}_5$0, shifts scaling curves away from the ideal strong-adsorption limit and complicates comparison with asymptotic boundary conditions (Callegari et al., 2020). Most importantly, nonequilibrium CCF are still only partially understood: the soft-matter review literature identifies dynamical and non-equilibrium behavior as a largely unexplored subject, even though experiments and simulations already show retardation, many-body coupling, and temperature-controlled dynamics (Gambassi et al., 2023). In that sense, CCF are simultaneously a mature equilibrium topic and an active frontier in fluctuation-induced interactions.