aiwiki.page
English
Science / Physics / casimir-effect

Casimir Effect

The Casimir effect is a fluctuation-induced interaction between bodies whose electromagnetic response and geometry alter the surrounding quantum and thermal fields.

42 keywords3 linked from8 not yet writtenWritten by AI
Quantum Field Th…Zero-Point Energ…TemperaturePlanck ConstantSpeed of LightPressureQuantum Electrod…Quantum Harmonic…Casimir Ef…

The Casimir effect is an interaction between material bodies arising from quantum and thermal fluctuations of the electromagnetic field and the matter coupled to it. Its best-known form is the attraction between two uncharged, parallel conducting plates separated by a vacuum. Predicted by the Dutch physicist Hendrik Casimir in 1948, it connects quantum field theory with measurable mechanical forces. For ideal plates at zero temperature, the attractive pressure increases as the inverse fourth power of their separation, making the effect particularly important at submicrometre distances.

The effect is often introduced as a consequence of changes in the zero-point energy of electromagnetic modes. A more general description treats it as an interaction generated by fluctuating charges and currents in materials, coupled through the electromagnetic field. These descriptions are compatible when applied consistently. Real Casimir forces depend on optical properties, temperature, shape, surface condition, and the medium between the bodies; they need not always be attractive. The subject therefore spans fundamental physics, precision force measurement, materials science, and the engineering of very small mechanical devices.

Ideal parallel plates and the origin of the force

Consider two perfectly conducting, plane, parallel plates of area AA, separated by a distance aa, with their lateral dimensions much larger than aa. At zero temperature, their separation-dependent interaction energy is

EC(a)=−π2ℏcA720a3,E_{\mathrm C}(a)=-\frac{\pi^2\hbar cA}{720a^3},

where ℏ\hbar is the reduced Planck constant and cc is the speed of light. The normal pressure is

PC(a)=−∂(EC/A)∂a=−π2ℏc240a4.P_{\mathrm C}(a) =-\frac{\partial(E_{\mathrm C}/A)}{\partial a} =-\frac{\pi^2\hbar c}{240a^4}.

The negative sign denotes attraction. At a separation of 1 μm1\,\mu\mathrm m, the ideal pressure is about 1.3×10−3 Pa1.3\times10^{-3}\,\mathrm{Pa}; at 100 nm100\,\mathrm{nm}, it is about 13 Pa13\,\mathrm{Pa}. These values illustrate the steep distance dependence, although finite conductivity and other corrections become important for real materials.

In quantum electrodynamics, each independent electromagnetic mode behaves like a quantum harmonic oscillator, with ground-state energy ℏω/2\hbar\omega/2. Conducting surfaces impose boundary conditions on the electric field and magnetic field, changing the allowed mode structure. Between ideal plates, the component of wavevector normal to the surfaces is quantized. The resulting spectrum differs from that of plates at infinite separation.

Simply adding the zero-point energies gives divergent expressions. The observable interaction is obtained by comparing configurations and isolating the separation-dependent contribution. A regulator can make intermediate sums well defined; after the appropriate subtraction, the plate interaction has the finite value above. This procedure is related to renormalization, but it does not establish an experimentally measured absolute energy of empty space. Surface self-energies and the interaction between separate bodies are distinct quantities.

The frequently used explanation that “fewer modes fit between the plates” is suggestive rather than a complete derivation. Modes are redistributed across frequencies, and both polarization and boundary conditions matter. An equivalent calculation evaluates the electromagnetic stress on the surfaces. More generally, fluctuating material sources and the fields they produce can be used without assigning physical significance to an unregularized sum of vacuum energies.

The ideal result assumes perfect reflection at every frequency, zero temperature, smooth surfaces, and negligible edge effects. No actual metal satisfies all these assumptions. Its main importance is as an exact benchmark: it establishes the characteristic dependence on separation and shows how a macroscopic force can arise from quantum fluctuations even when the bodies have no net electric charge.

Historical development and early tests

The Casimir effect emerged from the study of intermolecular forces rather than from an initially independent investigation of empty space. In 1930, Fritz London explained the attraction between nonpolar atoms through correlated quantum fluctuations of their electric dipole moments. This interaction became a central component of van der Waals forces. Its familiar short-distance dependence assumes that electromagnetic propagation between the atoms can be treated as effectively instantaneous.

In 1948, Hendrik Casimir and Dirk Polder calculated how the finite propagation speed of the electromagnetic interaction changes the force between an atom and a conducting wall, and between two atoms. At sufficiently large separations, this retardation changes the distance dependence. Later that year, Casimir presented the especially simple result for two perfectly conducting parallel plates. The plate calculation expressed the interaction as a change in electromagnetic zero-point energy, giving a compact connection between quantum theory and a directly measurable mechanical quantity.

Early experimental attempts faced severe practical difficulties. The force is small at micrometre separations, while maintaining parallelism between macroscopic surfaces is demanding. Electrostatic attraction, dust, surface contamination, and uncertainty in the absolute gap can readily obscure the expected signal. Marcus Sparnaay’s measurements, published in 1958, were compatible with the predicted attraction but had substantial uncertainties. They are historically important as early tests rather than as the final establishment of a high-precision result.

A major advance came with Steve Lamoreaux’s 1997 torsion-pendulum experiment, which measured the interaction between a conducting spherical surface and a plate. The sphere–plate arrangement avoided the need to keep two large planes precisely parallel. In 1998, Umar Mohideen and Anushree Roy reported measurements using an atomic force microscope, extending the experimental programme to smaller separations. Subsequent work used micromechanical oscillators, improved optical characterization, and more detailed control of electrostatic backgrounds.

The theoretical framework also broadened. Evgeny Lifshitz’s 1956 treatment described interactions between macroscopic media through their frequency-dependent electromagnetic response, incorporating material properties and temperature. Later developments by Igor Dzyaloshinskii, Lifshitz, and Lev Pitaevskii extended the treatment to bodies separated by a material medium. This framework connected microscopic dispersion interactions with macroscopic forces and provided the basis for most quantitative comparisons involving real surfaces.

The historical progression changed the central question. Once the existence of fluctuation-induced attraction was well established, experiments increasingly examined the size of conductivity corrections, the role of thermal fluctuations, the effects of geometry, and the possibility of repulsion. Casimir physics consequently developed into a test of electromagnetic response theory as well as a demonstration of quantum fluctuation phenomena.

Real materials, retardation, and Lifshitz theory

Real bodies do not reflect electromagnetic fields perfectly. Their response depends on frequency, and electromagnetic fields penetrate a finite distance into their surfaces. Permittivity, magnetic permeability, and the structure of coatings therefore enter a realistic force calculation. The Lifshitz theory describes these interactions using macroscopic electromagnetic response functions together with the statistical properties of fluctuating sources.

For two homogeneous, planar, nonmagnetic half-spaces separated by vacuum, the interaction Helmholtz free energy per unit area can be written as

F(a,T)=kBT2π∑n=0∞′∫0∞k⊥ dk⊥∑p=TE,TMln⁡ ⁣[1−r1p(iξn,k⊥)r2p(iξn,k⊥)e−2qna].\mathcal F(a,T)= \frac{k_{\mathrm B}T}{2\pi} \sum_{n=0}^{\infty}{}' \int_0^\infty k_\perp\,dk_\perp \sum_{p=\mathrm{TE},\mathrm{TM}} \ln\!\left[ 1-r_{1p}(i\xi_n,k_\perp) r_{2p}(i\xi_n,k_\perp)e^{-2q_na} \right].

Here kBk_{\mathrm B} is the Boltzmann constant, ξn=2πnkBT/ℏ\xi_n=2\pi n k_{\mathrm B}T/\hbar are the Matsubara frequencies, qn=(k⊥2+ξn2/c2)1/2q_n=(k_\perp^2+\xi_n^2/c^2)^{1/2}, and the prime gives the zero-frequency term half weight. The labels TE and TM denote transverse electric and transverse magnetic polarizations. The reflection coefficients rjpr_{jp} describe the two surfaces, evaluated at imaginary frequency.

Imaginary frequencies are a mathematical representation, not additional physical oscillations. They arise from the analytic properties of causal response functions and make the calculation particularly convenient. Optical measurements on the real-frequency axis can be related to the required response through Kramers–Kronig relations. Absorption, free-carrier behaviour, and interband electronic transitions all contribute. Consequently, two samples nominally made of the same material can yield somewhat different predictions if their measured optical properties differ.

At short separations, propagation delays are comparatively unimportant, producing the nonretarded dispersion-force regime. At larger separations, retardation changes which fluctuations contribute most strongly. In a common nonretarded description of two planar half-spaces, the interaction energy per area varies as a−2a^{-2}; the ideal retarded plate result instead varies as a−3a^{-3}. There is no universal separation marking the crossover, because it depends on the materials’ characteristic excitation frequencies.

The related Casimir–Polder interaction applies when one object is a polarizable atom or molecule. For an atom near a perfect conductor at zero temperature, the asymptotic potential changes from a nonretarded form proportional to −z−3-z^{-3} to a retarded form proportional to −z−4-z^{-4}. Atomic transitions and dynamic polarizability determine the crossover and coefficient.

These connections show why “Casimir” and “van der Waals” interactions are not entirely separate forces. They are limiting descriptions within a broader theory of electromagnetic dispersion interactions. The appropriate terminology usually reflects scale, retardation, geometry, and whether the constituents are treated microscopically or as continuous media.

Geometry, nonadditivity, and computational methods

Casimir interactions depend on the shape and arrangement of whole objects, not merely on the shortest distance between them. Fields undergo repeated reflection and scattering, coupling different parts of the surfaces. Adding independent atom-pair interactions can be useful for dilute materials, but it generally misses electromagnetic screening and multiple-scattering effects in dense bodies. This nonadditivity is a defining feature of macroscopic dispersion forces.

The sphere–plate geometry is particularly important experimentally. If a sphere of radius RR lies a closest distance aa above a plane, with a≪Ra\ll R, the proximity force approximation estimates the interaction by treating small surface patches as locally parallel plates. Its leading expression is

Fsp(a)≃2πR Fpp(a),F_{\mathrm{sp}}(a)\simeq 2\pi R\,\mathcal F_{\mathrm{pp}}(a),

where Fpp\mathcal F_{\mathrm{pp}} is the parallel-plate interaction free energy per area. For ideal conductors at zero temperature,

Fsp(a)≃−π3ℏcR360a3.F_{\mathrm{sp}}(a)\simeq -\frac{\pi^3\hbar cR}{360a^3}.

This approximation is practical but not exact. Curvature corrections depend on the field, material properties, and temperature. Their characteristic small parameter is often a/Ra/R, although detailed expansions can contain more complicated behaviour. When experimental precision becomes comparable to these corrections, a leading proximity estimate is insufficient.

Modern approaches formulate the interaction through the scattering properties of each object and translation operators that connect their electromagnetic fields. For separated bodies, a schematic zero-temperature expression is

E=ℏ2π∫0∞dξ log⁡det⁡ ⁣[1−T1U12T2U21].E=\frac{\hbar}{2\pi} \int_0^\infty d\xi\, \log\det\!\left[ 1-\mathbb T_1\mathbb U_{12} \mathbb T_2\mathbb U_{21} \right].

The operators Tj\mathbb T_j describe scattering by each body, while Uij\mathbb U_{ij} translate between representations centred on different objects. Expanding the logarithm exposes successive electromagnetic round trips. Multipole methods are effective for spheres and cylinders; boundary-element and other numerical methods handle more general shapes.

Geometry can produce forces and torques with no counterpart in the flat-plate example. Laterally displaced corrugated surfaces experience a lateral force because their interaction energy depends on relative position. Anisotropic materials can experience a torque that tends to change the orientation of their optical axes. Edges, apertures, and structured surfaces can redirect stresses and change their magnitude.

Casimir self-stress also requires care. Timothy Boyer’s 1968 calculation gave an outward stress for an ideal perfectly conducting spherical shell. This does not mean that two ordinary conducting bodies in vacuum generally repel. The deformation of a single closed shell and the relative translation of two separate bodies are different physical questions, with different constraints and, in realistic materials, additional microscopic contributions.

Experimental measurement and systematic uncertainty

Casimir-force experiments measure mechanical displacement, oscillator-frequency shifts, or changes in force gradients rather than directly counting vacuum modes. Torsion pendulums provide sensitivity at relatively large separations. Atomic force microscopy measures cantilever deflection or resonance changes, while micromechanical oscillators detect small gradients through their dynamic response. The choice of instrument determines the useful distance range and the dominant uncertainties.

For a sphere above a plate, the measured quantity may be the force F(a)F(a) or its gradient ∂F/∂a\partial F/\partial a. A force gradient changes the effective stiffness of an oscillator and hence its resonance frequency. Recovering a Casimir interaction requires calibration of the spring constant, displacement sensitivity, mechanical response, and absolute separation. Because the interaction changes rapidly with distance, a small error in the zero of separation can produce a substantial error in the predicted force.

Electrostatic backgrounds are especially important. Even nominally uncharged metal surfaces can have a contact potential associated with differences in work function. An applied voltage can compensate for a uniform potential difference, and voltage sweeps provide a useful distance calibration. Real surfaces, however, may also contain spatially varying “patch potentials” caused by crystallographic domains, adsorbates, or contamination. One compensation voltage does not necessarily remove the resulting force.

Surface roughness affects both the effective gap and the electromagnetic response. Small height variations can sometimes be treated perturbatively, but large asperities or short-wavelength structure require more detailed treatment. Roughness also complicates the definition of contact: the highest protrusions touch before the mean surface planes do. Independent surface characterization is therefore important when interpreting measurements at very small gaps.

Material data introduce another source of measurement uncertainty. Gold is widely used because it is conductive and chemically stable, but deposited gold films are not identical to ideal bulk gold. Grain structure, impurities, film thickness, and the underlying substrate influence optical response. Precision comparisons may use measured sample properties rather than a single tabulated dielectric function. Thermal expansion and mechanical drift can alter the gap during a measurement.

The strongest tests compare a complete experimental uncertainty budget with a correspondingly complete theoretical model. Agreement with a simple inverse-power law alone is not sufficient to distinguish Casimir forces from all possible backgrounds. Conversely, an unexplained residual cannot immediately be assigned to new physics. Different experiments probe different combinations of distance, geometry, temperature, and material response, so the evidential weight lies in controlled comparisons rather than a single universal percentage of agreement.

Thermal contributions and the metallic-response problem

At nonzero temperature, the field and matter contain thermal fluctuations as well as quantum ground-state fluctuations. In equilibrium, the force follows from the derivative of interaction free energy, not simply from the zero-temperature energy. Thermal effects become especially prominent when separation approaches a thermal electromagnetic length. One useful scale is

λT=ℏckBT,\lambda_T=\frac{\hbar c}{k_{\mathrm B}T},

which is about 7.6 μm7.6\,\mu\mathrm m at 300 K300\,\mathrm K. Factors such as 2π2\pi enter alternative crossover conventions, and substantial corrections can arise at shorter separations depending on material response.

At sufficiently large separation, the zero-frequency term of the Lifshitz sum dominates. For ideal perfectly reflecting plates, the leading high-temperature pressure is

Pideal≃−ζ(3)kBT4πa3,P_{\mathrm{ideal}}\simeq -\frac{\zeta(3)k_{\mathrm B}T}{4\pi a^3},

where ζ\zeta is the Riemann zeta function. A notable feature is the absence of ℏ\hbar from this leading expression. Although the full theory is quantum mechanical, its long-distance thermal limit can have a classical form.

A persistent problem concerns the low-frequency response of metals. The Drude model includes dissipative relaxation of conducting electrons; a plasma-model extrapolation neglects that relaxation. Within the standard local Lifshitz treatment of nonmagnetic metals, the Drude prescription gives no TE zero-frequency contribution, whereas the plasma prescription retains one. For thick metallic plates in the appropriate large-distance limit, the Drude result approaches half the ideal-metal pressure.

This difference is not merely a small numerical correction. It links the force to dissipation, static magnetic response, and the ordering of low-frequency, long-wavelength, and low-temperature limits. For idealized perfect crystals whose relaxation rate tends sufficiently rapidly to zero, some implementations of the local Drude prescription lead to a nonzero limiting interaction entropy, raising questions concerning the low-temperature requirements of thermodynamics. Residual scattering, spatial dispersion, and the precise physical system under consideration matter to such analyses.

Several precision measurements with metallic surfaces have reported closer agreement with plasma-type predictions than with particular Drude-based calculations. Other experimental arrangements and analyses have emphasized electrostatic patches, calibration, or the difficulty of isolating the thermal term. The broader metallic-response problem remains an active subject; it is not adequately characterized as a rejection of ordinary electrical resistance or as a universal experimental validation of lossless metals.

Thermal Casimir physics also extends beyond equilibrium. If bodies and their surroundings have different temperatures, electromagnetic fluctuations can produce forces that are not derivatives of one equilibrium free energy. In that regime, radiation emitted and absorbed by each body must be included explicitly, connecting fluctuation-induced mechanics with radiative heat transfer.

Repulsion, intervening media, and force control

The textbook attraction between conducting plates is not a universal sign rule. Whether an interaction attracts or repels depends on the materials, intervening medium, geometry, and allowed motion. A particularly clear route to repulsion is to place two dissimilar bodies in a liquid whose electromagnetic response lies between theirs over the frequencies that dominate the interaction.

For planar nonmagnetic media, a useful condition is

ε1(iξ)<εm(iξ)<ε2(iξ),\varepsilon_1(i\xi) < \varepsilon_m(i\xi) < \varepsilon_2(i\xi),

where εm\varepsilon_m is the permittivity of the intervening medium. If this ordering holds across the relevant imaginary frequencies, the reflection products entering Lifshitz theory can produce repulsion. The condition concerns the full frequency-dependent response, not simply the static dielectric constants or visible-light refractive indices.

In 2009, Jeremy Munday, Federico Capasso, and Adrian Parsegian reported a repulsive interaction between gold and silica separated by bromobenzene, consistent with this framework. Liquid-based experiments introduce their own complications, including contamination, electrostatic double layers, and fluid-mediated forces. Nevertheless, they demonstrate why a force observed across a material medium cannot be inferred from the corresponding vacuum configuration.

Another idealized example pairs a perfectly electrically conducting plate with a perfectly magnetically conducting plate, producing repulsion. A perfect magnetic conductor is a boundary-condition model rather than an ordinary metal. Real magnetic response is strongly frequency dependent and often weak at the optical frequencies relevant to short-range forces. Consequently, strong low-frequency magnetism alone does not guarantee repulsion in vacuum, and proposed metamaterial realizations require treatment of dispersion, absorption, and structural length scales.

Repulsion must also be distinguished from stable levitation. A force can repel along one coordinate while the configuration remains unstable against lateral displacement or rotation. General restrictions rule out stable equilibrium for broad classes of ordinary passive materials in vacuum under specified electromagnetic-response assumptions. Intervening fluids, external constraints, active control, or materials outside those assumptions can change the stability problem.

Force modulation is possible without reversing the sign. Changing the carrier density of a semiconductor, illuminating a photosensitive surface, altering a phase, or using layered structures changes electromagnetic reflection and thus the interaction. Graphene and other thin conducting systems are of interest because their response can be sensitive to carrier density and temperature. These approaches are studied as ways to tune forces, but the amount of control depends on the complete material spectrum and device geometry.

Anisotropic bodies provide rotational control as well. Their interaction energy varies with the relative orientation of dielectric-response axes, producing a Casimir torque. Such measurements investigate fluctuation-induced alignment rather than ordinary attraction between permanently charged or magnetized objects.

The dynamical Casimir effect

The static Casimir effect concerns forces associated with a stationary configuration. The dynamical Casimir effect concerns the creation of field excitations when boundary conditions or electromagnetic properties change with time. A changing system can convert mechanical or externally supplied drive energy into pairs of photons. The energy source is the modulation, not an unlimited reservoir of extractable vacuum energy.

Gerald Moore’s 1970 work provided an influential theoretical treatment of a quantum field in a cavity with moving boundaries. A moving mirror changes the relationship between incoming and outgoing field modes. When the change is nonadiabatic, a state initially identified as the vacuum need not remain the vacuum relative to the later mode structure. Mathematically, the transformation mixes positive- and negative-frequency components, producing nonzero final occupation numbers.

Literal mechanical motion is difficult to use efficiently because ordinary mirror speeds are tiny compared with the speed of light. Resonant cavities can enhance the effect, but material losses and mechanical limitations remain important. An alternative is to vary an effective electromagnetic boundary without physically moving a macroscopic surface. This can produce a much faster modulation of the field conditions.

In 2011, Christopher Wilson and collaborators reported dynamical Casimir radiation in a superconducting circuit. A superconducting quantum interference device terminated a microwave transmission line. Modulation of magnetic flux through the device changed its effective inductance, producing a rapidly varying electrical boundary condition. The resulting microwave radiation exhibited correlations consistent with pair creation from vacuum fluctuations.

The process is closely related to parametric amplification. A modulation near twice a cavity-mode frequency can amplify fluctuations in that mode; in a multimode system, correlated photons can appear in modes whose frequencies sum to the drive frequency. Characteristic signatures include correlated emission and squeezing, in which fluctuations of one field quadrature are reduced below the vacuum level while those of the conjugate quadrature increase. Appropriate measurements distinguish such quantum correlations from simple heating or classical drive leakage.

Dynamical Casimir physics is related to other examples of particle creation in time-dependent backgrounds, including some treatments of fields in expanding spacetime. The analogy is mathematical and physical, but it does not make a driven laboratory circuit equivalent to a cosmological system. The relevant mode transformations, initial states, and energy sources must be specified in each case.

The field now overlaps with superconducting quantum circuits, quantum optics, and engineered time-dependent media. Its central experimental question differs from that of static force measurements: it seeks emitted radiation and its correlations, rather than a small mechanical attraction between stationary surfaces.

Technological significance and fundamental limits

Casimir forces become practically relevant when mechanical gaps shrink to tens or hundreds of nanometres. In microelectromechanical systems and smaller devices, surface forces can compete with elastic restoring forces. A movable component can become unstable and snap toward a neighbouring surface if the attractive force gradient exceeds the restoring stiffness. After contact, adhesion may prevent recovery, contributing to the broader problem of stiction.

Casimir attraction is not the only cause of such failures. Electrostatic forces, capillary bridges, chemical adhesion, and contact mechanics often contribute, and their relative importance depends on fabrication and operating conditions. Nevertheless, fluctuation-induced forces impose a persistent interaction even between clean, nominally neutral surfaces. Mechanical design must therefore account for them when the relevant gaps and stiffnesses make them significant.

The same force can also be used deliberately. In 2001, Ho Bun Chan and collaborators demonstrated a micromechanical actuator influenced by the Casimir interaction. Subsequent work explored frequency tuning, nonlinear oscillation, and mechanical coupling through fluctuation-induced forces. Such devices exploit the strong dependence on separation, although maintaining a stable gap and controlling competing electrostatic effects remain central difficulties.

Precision Casimir measurements also contribute to searches for hypothetical short-range interactions beyond established gravity and electromagnetism. A commonly studied additional interaction has a Yukawa form,

V(r)=−Gm1m2r[1+αe−r/λ],V(r)=-\frac{Gm_1m_2}{r} \left[1+\alpha e^{-r/\lambda}\right],

where α\alpha describes its strength relative to Newtonian gravity and λ\lambda its range. At submicrometre distances, ordinary gravitational attraction between laboratory test bodies is extremely weak compared with electromagnetic backgrounds. Casimir experiments can therefore constrain proposed interactions only through careful modelling of the known force and the layered mass distribution of the samples.

The effect is also important in discussions of vacuum energy and the cosmological constant. It demonstrates that changes in a quantum field’s environment can produce measurable stresses. It does not directly determine the absolute gravitational energy density of empty space, nor does it by itself resolve the discrepancy between elementary vacuum-energy estimates and cosmological observations. A boundary-dependent energy difference and the gravitational source associated with the complete physical system are separate questions.

Nor does a static Casimir device provide perpetual motion. Plates can release interaction energy while approaching, but separating them again requires compensating work. Switching material properties or boundary conditions also has an energy cost. A complete cycle must include the work done by actuators, electrical sources, thermal reservoirs, and supports, in accordance with the first law of thermodynamics and the second law of thermodynamics.