Casimir Physics and Vacuum-Force Engineering

The last chapter said the vacuum is a structured medium. This one hands you the tools. There is exactly one place where a laboratory reaches out and touches the zero-point field with an ordinary instrument, and it is the gap between two close-set plates. Everything downstream in this book — power cells, tunnelling diodes, torque elements, the negative-energy budget of Part V — is engineering built on three pages Hendrik Casimir read to the Royal Netherlands Academy on 29 May 1948.

The prediction, in one move

Casimir's argument is startlingly plain, and you can follow it in a paragraph. The electromagnetic field has a lowest state that still jitters. Put a conducting plate inside a conducting box and count the field modes that fit when the plate sits close to one wall, then count them again when it sits far away. Both totals are infinite and, as Casimir himself says, meaningless on their own. The difference between them is finite, and it depends on the size of the gap. Differentiate that difference with respect to the gap and you have a force per unit area.

Casimir pressure between two ideal mirrors(15.1)

$$\dfrac{F}{A} = -\dfrac{\pi^2 \hbar c}{240\, a^4}$$

Casimir's own name for what he had found was a zero-point pressure of electromagnetic waves, which is why this chapter sits in Part II rather than in a curiosities appendix. Read the original three pages and there is no exotic ingredient anywhere in the derivation — just mode counting, and a subtraction that leaves something finite behind.

Measuring it twice, in different laboratories

The prediction sat almost untested for fifty years. In 1958 M. J. Sparnaay at Philips built the first apparatus that went looking, reported an attraction consistent with Casimir's law, and — this is the part that earns the paper its place — named the artefact that had wrecked his earlier runs: a stray voltage difference between the two facing surfaces, pulling them together electrically. Every precision Casimir experiment since has had to cancel that voltage before it can see the vacuum at all.

Then two measurements arrived a year apart, by different methods, in different laboratories. Steve Lamoreaux hung a torsion pendulum in a vacuum chamber, gave up on parallel plates because holding two centimetre-wide plates parallel to ten microradians is not practical, and used a gold-coated lens against a flat instead. Across gaps from 0.6 to 6 micrometres the force matched theory to about five per cent with no adjustable parameters. The following year Umar Mohideen and Anushree Roy glued a 196-micrometre sphere to an atomic-force cantilever and walked it in from 0.9 micrometres to 0.1, recording the force at 256 separations; the agreement is about one per cent at closest approach, sharp enough to resolve the corrections for finite conductivity and surface roughness that the bare formula leaves out.

Definitive Two independent instruments, two laboratories, one answer. That is what carries a claim to settled physics on this site's ladder, and the Casimir force is the cleanest example the whole book contains.

The record only got sharper. In 2002 an Italian group did it Casimir's own way, holding a chromium-coated cantilever parallel to a rigid surface to better than thirty nanometres across a millimetre and reading the force off the shift in the cantilever's resonant frequency. In 2007 a micromechanical torsional oscillator reached a total error of 0.19 per cent at a 160-nanometre gap — the first time in this field that random scatter fell below systematic error. And in 2011 a torsion pendulum separated the force into its two pieces by distance, measuring the thermal Casimir force between macroscopic bodies for the first time: zero-point fluctuations dominate close in, room-temperature thermal fluctuations dominate beyond about three micrometres and fall off more slowly. That result is also the best practical reminder of what the word zero means here. It means zero thermal temperature, and nothing else — pull the plates far enough apart in a warm room and the thermal part of the field is the larger of the two.

Real materials are not ideal mirrors

The moment you build something, formula 15.1 stops being enough. A real mirror reflects well only below its plasma frequency, so at a few hundred nanometres the finite conductivity of gold or aluminium matters; so does roughness; so does temperature. The practitioner's tool for all of it is scattering theory: describe each surface by how it reflects light at every frequency, and the force follows from those reflection amplitudes alone. The standard treatment also delivers a warning worth carrying — the popular shortcut of treating a bumpy surface as a stack of flat ones underestimates the roughness correction by about sixty per cent at a two-hundred-nanometre gap.

The decade in which this became metrology is audited in The Casimir force between real materials and in Advances in the Casimir Effect, the field's standard book: several techniques, several continents, agreement at the level of parts per thousand — plus a genuine live dispute the authors name rather than smooth over, about how the conduction electrons of a real metal should enter the temperature term. A materials-side companion, Materials perspective on Casimir and van der Waals interactions, makes the engineering point that changes everything: the gecko force and the mirror force are one phenomenon read at different distances, and graphene, topological insulators and metamaterials give it a response no ordinary metal offers. The force has become a design variable.

Geometry is the other design variable

Change the shape and you change the law, not just the number. An exact calculation for a parabolic mirror sharpened to a knife's edge shows the standard rule of thumb predicting no force at all where the true energy falls off as one over the gap squared. Push two spheres far enough apart and the force forgets the material entirely, becoming universal in temperature and shape — a regime unreachable in vacuum but reachable at a hundred nanometres in salt water, where optical tweezers can hold the beads.

Then the experiments caught up with the theory. A Hong Kong and Princeton collaboration etched two facing arrays of T-shaped protrusions out of a single silicon chip so that alignment came free with the lithography, and measured a force gradient that changes sign twice as the gap closes — at one position the vacuum stiffens the beam, a spring made of fluctuations. A Davis group held the material fixed and varied only the shape, and found that a single 300-nanometre pillar cuts the force roughly tenfold while a dense array nearly triples it, in ordinary air at room temperature, with plain theory close enough to design from.

For shapes nobody has solved on paper, there are numbers. Finite-difference computation runs ordinary classical electromagnetism software on Casimir problems in any geometry. Worldline numerics — a loop-cloud method — was used under a DARPA grant to model a Casimir cavity with pillars standing on the midplane, and the team found the pillar does not shield itself but strengthens the field around it by a factor of three to five; then they noticed that the computed energy distribution has the shape Alcubierre's warp metric asks for. That is a lead, not a warp drive, and it is filed here because it is first of all a Casimir calculation with a chip you could fabricate at the end of it.

Making the vacuum push

For sixty years the force only pulled, and a theorem said mirror-image bodies could never do otherwise. The theorem is true and its assumptions are escapable, and getting out of it is now a research programme with four families in it: put a liquid in the gap, use magnetic materials, drive the system out of equilibrium, or sculpt the surfaces. The survey of all four weighs each by how close it is to a working experiment, and names the prize: a levitating component with almost no friction at all.

One family has already been measured. Lifshitz theory says the sign of the force follows the ordering of the three optical responses.

The ordering that reverses the sign of the force(15.2)

$$\varepsilon_1(i\xi) \;>\; \varepsilon_{\text{fluid}}(i\xi) \;>\; \varepsilon_2(i\xi)$$

Gold sphere, silica plate, bromobenzene in between: the cantilever bent away instead of toward, on approach and on retraction alike, and repulsion was on the record. Since then the routes have multiplied. Pair a mirror with a strongly magnetic film and the sign flips, with a simple design rule for how magnetic the film has to be. Put a modest forward bias across a semiconductor and its emitted photons carry a chemical potential set by the applied voltage, so that gallium arsenide facing gold turns repulsive beyond a few hundred nanometres — a sign you can switch with a power supply while the device is running. Fill the gap with a chiral medium and the theorem's key step fails outright, giving a force that oscillates between attraction and repulsion and tunes with a magnet. Or use a ferrofluid, where four independent dials — the metal, a Teflon thickness, the nanoparticle size and the solvent — place a stable trap at a fixed spacing held open by the vacuum alone.

Strong Repulsive vacuum forces are measured in fluids and designed, not yet built in several of the vacuum configurations. The measurement that would settle the vacuum case is a sphere-and-plate run showing tens to hundreds of piconewtons of push in a chamber, using apparatus every Casimir laboratory already owns.

The vacuum also twists

Give the plates a direction — a birefringent crystal, whose optical response depends on which way its axis points — and the field does not merely pull them together, it tries to rotate them into alignment. The review of the Casimir torque is mostly about apparatus, because the trick that saves the force experiments, swapping a plate for a sphere, does not work when the material has an axis. It offers three benches: a disc floating on a repulsive force, a three-threaded pendulum, a levitated nanorod. Adding the thermal part of the spectrum shows it fighting the quantum part rather than joining it — barium titanate at room temperature loses about 99.5 per cent of its torque, where the same correction trims the ordinary force between metals by about a third. Cool the apparatus and the torque grows while the noise shrinks. And a 2026 proposal uses altermagnets, whose spin pattern is tied to the crystal's own rotation symmetry, to switch the twist on with a field pointing straight through the plates, with the torque climbing as the square of the field and its direction chosen by an underlay.

From a force to a machine

This is where the chapter earns its title. In 2001 a Bell Laboratories team hung a doped polysilicon plate on two thin torsion rods, brought a metallised sphere up to one side, and watched the plate rotate: the first micromechanical device driven by vacuum fluctuations. The engineering lesson landed with it — once parts sit nanometres apart, quantum electrodynamics is part of the mechanical design and not a footnote to it. A second line rebuilt a 1972 capacitive detection scheme around modern nanopositioners to demonstrate actuation of a membrane by the Casimir force, spending most of its effort on the electrostatic calibration you must subtract before you may claim a force at all.

Then the force stopped being the signal and became the tool. A Kassel group patterned metal shutter blades on a chip, released them, and let a drying step bring neighbours close enough that Casimir and van der Waals forces snapped them into interlocking pairs — structure assembled out of a flat chip by the vacuum, with the hard part being prising the pairs open again. The same author who built the membrane rig argues in a keynote on nanotechnology that a semiconductor's Casimir force changes under illumination, which means a laboratory knob moves a vacuum force in real time — and a force you can switch is a force you can drive around a loop. His aerospace metrology talk places those examples on the readiness matrix engineers use to judge how near a technology is to flight.

Two further benches are worth knowing by name. Archimedes, funded by Italy's INFN, would cycle high-temperature superconducting plates through their transition so that a cavity's reflectivity — and therefore its vacuum energy — switches on and off, and weigh the difference on a balance at around ten to the minus sixteen newtons. It is an attempt to answer whether the vacuum's own energy falls in a gravitational field. And in 2026 a group laid a two-layer flake of niobium diselenide under an unpowered terahertz antenna whose only job is to reshape the electromagnetic vacuum around it; the superconducting transition rose from 3.02 to 3.41 kelvin, tracking the antenna's simulated field map along the flake and reversing when the antenna was detuned. Shape the vacuum with hardware and matter answers.

The honest interpretation point

Here is the part of this chapter that makes it the best teaching case on the site. In 2005 Robert Jaffe showed that the same force comes out a second way — from ordinary quantum electrodynamics, by summing diagrams for the charges and currents in the plates themselves, with no appeal to vacuum fluctuations at all. Read that way the Casimir force is the relativistic, retarded van der Waals force, and it fades as the fine-structure constant is turned down. Charles Chase states this against his own programme on air; Ashton Forbes accepts it on air. And the second clause matters as much as the first, and is usually dropped: quantum electrodynamics still requires the vacuum fluctuation term for the mathematics to work, and Jaffe says plainly that he expects zero-point energies to stay in the theory.

Definitive The force is measured. Suggestive The interpretation is underdetermined. Strong The engineering works either way, because an actuator does not care which calculation you used to design it. Say all three sentences. A claim stated at exactly its own strength cannot be taken away from you.

Who builds this

This unit feeds trades, not just theories: MEMS and NEMS process engineers, precision-metrology engineers, semiconductor and foundry process engineers, surface and thin-film physicists, atomic-force-microscopy specialists, nanofabrication technicians. If you would rather do vacuum engineering in a cleanroom than at a blackboard, this is the door, and the skills are the ones a chip foundry hires for today.

What the field added — July to September 2026

The season's clearest gain was the cavity-superconductor result: a resonator with no power supplied to it, reshaping the vacuum around a flake and raising its transition temperature by about ten per cent, with a second group reporting a comparable shift on a bilayer in a separate resonator. Two apparatus, two teams, not two versions of one paper. The Casimir force showed that the vacuum pushes on plates; this shows the vacuum changing a material's quantum state.

The season also sharpened the device conversation. Harold White's Casimir power cell was described on air in enough detail to draw: a static cavity, two walls, three pillars standing in the middle to break the symmetry, the pillars electrically isolated so that electrons tunnel across to them and the tunnelling current is the output. Every device in Part III pulls the levers this chapter has just described — material, geometry, boundary motion — which is why the honest place to learn them is here, where the force is measured to a part in five hundred.

---

Where each claim stands

| Claim | Maturity | What would settle it | |---|---|---| | Two uncharged plates attract in vacuum, following Casimir's law | Settled physics | Already settled: two independent instruments in 1997 and 1998, sub-percent since | | The force depends on material, roughness and temperature in calculable ways | Settled physics, with one named live dispute | The temperature term for real metals, decided by a measurement that separates the Drude and plasma descriptions | | Geometry alone changes the force tenfold down or threefold up, and can change its sign | Published and peer-reviewed, measured on chips | Independent replication of the non-monotonic silicon result in a second laboratory | | Repulsive Casimir forces in a fluid | Settled physics, measured 2009 | Already measured; the open case is repulsion in vacuum | | Repulsion in vacuum from magnetic films, biased semiconductors, chiral media or ferrofluids | Designed, not yet built | A sphere-and-plate run reporting tens to hundreds of piconewtons of push | | The Casimir torque as a switchable, engineerable effect | Published and peer-reviewed for the theory; on the bench now for the apparatus | A second measurement of the torque, and a demonstration that a field switches it | | The Casimir force proves an extractable zero-point reservoir | What to watch | An extraction device that closes its own energy ledger; the force alone does not decide it |

Sources

The one place in this book where the vacuum is weighed on a bench, with the interpretation dispute kept in plain view.

Primary — the prediction and the measurements

• H. B. G. Casimir (1948), "On the attraction between two perfectly conducting plates," Proc. Kon. Ned. Akad. Wetensch. B51, 793. /library/stm-11433059a4.
• M. J. Sparnaay (1958), "Measurements of attractive forces between flat plates," Physica 24, 751 — the first measurement, and the first named systematic. /library/stm-2aa45438b3.
• S. K. Lamoreaux (1997), "Demonstration of the Casimir force in the 0.6 to 6 micrometre range," Phys. Rev. Lett. 78, 5. /library/stm-e8c853485e.
• U. Mohideen & A. Roy (1998), "Precision measurement of the Casimir force," Phys. Rev. Lett. 81, 4549. /library/stm-911b036fc2.
• G. Bressi, G. Carugno, R. Onofrio & G. Ruoso (2002), parallel-plate measurement. /library/stm-208d347532.
• R. S. Decca et al. (2007), Casimir pressure to 0.19 per cent. /library/stm-3241e37fea.
• A. O. Sushkov, W. J. Kim, D. A. R. Dalvit & S. K. Lamoreaux (2011), "Observation of the thermal Casimir force." /library/stm-4fb469cffa.

Theory for real boundaries

• A. Lambrecht, P. A. Maia Neto & S. Reynaud (2006), "The Casimir effect within scattering theory." /library/stm-eb46d8bcd7.
• M. Bordag, G. L. Klimchitskaya, U. Mohideen & V. M. Mostepanenko (2009), Advances in the Casimir Effect. /library/stm-30ce9fddc2.
• G. L. Klimchitskaya, U. Mohideen & V. M. Mostepanenko (2009), "The Casimir force between real materials." /library/stm-7e33aa610f.
• K. A. Milton (2004), "The Casimir effect: recent controversies and progress." /library/stm-719d2e0a0b.
• S. K. Lamoreaux (2005), "The Casimir force: background, experiments, and applications." /library/stm-4c6743a6b7.
• A. W. Rodriguez, F. Capasso & S. G. Johnson (2011), "The Casimir effect in microstructured geometries." /library/stm-d41136300f.

The interpretation dispute, stated fairly

• R. L. Jaffe (2005), "The Casimir effect and the quantum vacuum," Phys. Rev. D 72, 021301 (arXiv:hep-th/0503158) — the force from radiation reaction, and the author's own statement of what he has not shown. /library/stm-dc5bf2b401.
• R. Onofrio (2006), "Casimir forces and non-Newtonian gravitation" — how accurately the force is known, and what that precision buys. /library/stm-8a6a8e0653.

Devices and benches

• H. B. Chan, V. A. Aksyuk, R. N. Kleiman, D. J. Bishop & F. Capasso (2001), Casimir actuation of a MEMS torsional device. /library/stm-50bc48707d.
• H. White, J. Vera, A. Han, A. R. Bruccoleri & J. MacArthur (2021), "Worldline numerics applied to custom Casimir geometry." /library/stm-1775ebeff1.
• E. Calloni et al. (2015), "The Archimedes experiment." /library/stm-f81b04508e.
• H. Zhang et al. (2026), "Cavity-enhanced superconductivity in the two-dimensional limit of NbSe2." /library/stm-812175a230.