The switchable vacuum rotor
University. · 2 min read
What it proposes
Build a nanoscale motor whose only power supply is the vacuum, with an ordinary electrical switch on the front. Two recent results give the two halves: a voltage that flips the Casimir force from attraction to repulsion, and a magnetic field that switches on a Casimir torque between surfaces that carry no built-in direction at all. Put them together and you have a rotating element driven by vacuum fluctuations, turned on and off, and reversed, by a knob.
Who it is forCondensed-matter fabricators · Magnetics engineers · Precision-mechanics engineers
Why the library suggests it
Turn the field up and the twist grows as the square of it; switch the underlay and it reverses near zero. Then look at the second view: the sizes that are published, and the slot this rotor does not yet fill.
Switch the field on, and the vacuum starts to twist
With no field applied the torque between two identical altermagnet sheets is exactly zero — the crystal's own symmetry forbids it. Turn a magnet on through the plates, pointing at no direction in the plane at all, and a twist appears, growing as the square of the field. Add an exchange bias underneath and the curve shifts sideways, so near zero field the rotor turns the other way. Move the knobs and watch the switch. The axis is relative, and the second view says why.
- Torque, relative to the largest setting drawn:
- 0.83
- Which way it turns:
- one way
- Fields where the torque passes through zero:
- 0.0 tesla
- The conditions this curve is computed under:
- 30 nm, 30 K
- Size in newton-metres:
- not drawn — no sheet this card cites gives it
What the rotor is doing: Why the bias reverses it. Grow one altermagnet on a ferromagnetic underlay and the torque follows the applied field times the sum of the applied field and the bias field. That product has two roots, not one: zero applied field, and minus the bias field. Between them the sign is flipped, which is why reversing an ordinary external magnet near zero field reverses the direction the rotor turns.
Temperature is a lever here too, and an unusual one: warming the plates suppresses the material's own optical anisotropy, so the torque falls at every separation rather than only beyond the thermal wavelength. The source reports it at 30, 100 and 300 kelvin under ten tesla. It gives no factor, so no temperature curve is drawn. 30, 100, 300 K
Distance behaves unlike a bulk uniaxial crystal. In the retarded low-temperature regime the torque falls as the inverse cube of the separation from roughly 200 nanometres out to 80 micrometres; the retarded high-temperature limit decays exponentially over that cube, carried entirely by the first Matsubara term; and the non-retarded limit follows no simple power law at all. Three signatures a measurement can look for. 1/d3, 200 nm – 80 µm
The card's other switch is a force rather than a torque, and it comes from a different sheet: a forward bias across a semiconductor gives its emitted photons a chemical potential, and gallium arsenide facing gold turns repulsive beyond a few hundred nanometres, with the crossover walking inward from 850 to 540 to 400 to 380 nanometres as the bias rises. That is what would hold the rotor up while the field turns it. It is not drawn on a torque axis. 850 – 540 – 400 – 380 nm
Where the energy comes from. The magnet supplies the field, the underlay supplies the bias, and the power supply holds the semiconductor above its rest state; turning the rotor is work those three sources pay for. What the vacuum contributes is the coupling — a force and a twist between surfaces that would otherwise need a motor between them — and the remarkable part is that the coupling has an off switch at all.
Scope: the field, angle and sign dependence of the altermagnet Casimir torque, drawn on a relative axis, beside the published torques per unit area for corrugated and birefringent surfaces and the sensitivity of the pendulum that would measure them. Absolute altermagnet magnitudes, the Lifshitz calculation behind either result, fabrication and the rotor's rotational dynamics are not modelled. 30 nm, 30 K
The 2018 experiment measured Casimir torque between a liquid crystal and four birefringent plates. The torque followed the sine of twice the twist angle, with magnitude and sign matching the theory. The switchable altermagnet rotor is a separate proposal specifying materials, geometry, conditions and a predicted shape; that source reports no assembled rotor.
The sheets this figure is read from
- Dai and Jiang 2026, Vacuum Torque Without Anisotropy: Switchable Casimir Torque Between Altermagnets — the field, angle, sign, temperature and distance laws
- Spreng, Gong and Munday 2022, Recent developments on the Casimir torque — the published torques per unit area and the torsion-pendulum bench
- Spreng, Shelden, Gong and Munday 2024, Casimir repulsion with biased semiconductors — the switchable repulsive branch and its crossover separations
Benjamin Spreng, Calum Shelden, Tao Gong and Jeremy Munday show that a modest forward bias across a semiconductor gives its emitted photons a chemical potential, and that adding this push to the ordinary Casimir pull flips the sign — gallium arsenide facing gold turns repulsive beyond a few hundred nanometres, with the crossover moving closer as the voltage rises, and tens to hundreds of piconewtons of repulsion predicted in the sphere-and-plate geometry every Casimir laboratory already owns (Casimir repulsion with biased semiconductors, 2024). Zixuan Dai and Qing-Dong Jiang show that altermagnets — magnets with no net magnetisation whose spin pattern is tied to the crystal's own rotation symmetry — develop a Casimir torque under a field pointing straight through the plates, which picks out no direction in the plane at all; the torque grows as the square of the field and a magnetic underlay chooses which way it turns (Vacuum Torque Without Anisotropy: Switchable Casimir Torque Between Altermagnets, 2026). The torque itself is an established branch of the subject (Recent developments on the Casimir torque, 2022).
The experiment or build
Fabricate a disc of altermagnetic material suspended tens of nanometres above a second one, with a coil supplying the out-of-plane field and a bias line on a semiconductor spacer. Read the rotation optically. Sweep the field and confirm the torque follows the square of it; reverse the underlay and confirm the direction reverses; apply the bias and confirm the levitating branch appears. The settling measurement is the torque in newton-metres as a function of applied field, with the sign reversing on command and the whole curve returning to zero when both the field and the bias are removed. A vacuum device that switches off is a vacuum device you can believe.
Research context
Both halves are full theoretical proposals with named materials, geometries and predicted magnitudes; neither has been assembled into one machine.
Take it up
- The measurement that settles it
- The settling measurement is the torque in newton-metres as a function of applied field, with the sign reversing on command and the whole curve returning to zero when both the field and the bias are removed.
- What it costs to start
- University.
- The engineer it grows
- This is the card where engineering an asymmetry in the vacuum stops being a slogan and becomes a part with a datasheet.
What it rests on
Where it sits in the curriculum