Ion-Wind Aircraft

1 One thruster stage

Cross-section of one emitter/collector pair, slowed down enormously. Grey dots are neutral air molecules; blue dots are positive ions born in the corona around the wire. Watch the grey dots get shoved leftward every time an ion passes.

Voltage across the gap Gap between wire and collector

Field across the gap

0onset 0.2MIT 0.8sparkover 1.2

Why a thin wire? The field is strongest where a conductor is sharpest. A 0.2 mm wire concentrates 20 kV into a field that ionises air within a millimetre of its surface, while the fat rounded collector never does.

Why does the air move slowly when the ions move fast? An ion drifts at 100–200 m/s but collides with a neutral molecule every few tens of nanometres. Ions are outnumbered about ten billion to one, so all that momentum is shared out across the whole air mass, which ends up moving at a few metres per second.

The efficiency trade. Thrust per watt equals gap ÷ (mobility × voltage), which is 1 ÷ (mobility × field). Wider gaps and lower voltages are more efficient but produce less thrust per metre of wire, so you need more wire. Push the voltage too far for the gap and the corona turns into an arc: see section 6.

2 The airplane: point the wind backwards

Side view, not to scale. Four rows of emitter wire and collector airfoil hang under the wing, each row running the full 5 m span. Below: the plane crossing the 60 m gym in real time.

3 The drone: point the wind down

Side view. The upper grid is a mesh of emitter wires, the lower grid a mesh of collectors. Air is pulled in from above the frame, accelerated between the grids, and thrown at the floor.

4 How much wire would it take?

Airplane (lift-to-drag 10)

m of emitter wire

Drone (thrust = weight)

m of emitter wire

5 Where do 40,000 volts come from?

The three-stage supply MIT flew: battery → high-frequency inverter → small step-up transformer → Cockcroft–Walton multiplier ladder. Dots are charge; watch packets climb the ladder one rung per half-cycle. A mirrored ladder with its diodes turned around makes the −20 kV side.

Not one big transformer. A transformer that turns 200 V into 40 kV directly needs a 1 : 200 winding ratio and insulation good for 40 kV, which means heavy iron and thick potting. The trick is to split the job.

Step 1: chop the DC into fast AC. Transistors switch the 200 V battery current on and off hundreds of thousands of times a second. A transformer's size shrinks as the frequency rises, so at this frequency a step-up transformer to a few kilovolts weighs grams, not kilograms.

Step 2: the multiplier ladder. A Cockcroft–Walton ladder is just diodes and capacitors. Each stage charges a capacitor to the AC peak and stacks it on the one below, so every rung adds twice the transformer's peak voltage. Five rungs of 4 kV make 20 kV. Old cathode-ray TVs, bug zappers and the hobby "lifters" all use the same ladder.

Why it stays light. The thrusters draw only about 13 mA, so every wire can be hair-thin. The real design driver is insulation: air breaks down at about 3 kV per millimetre, so 40 kV can jump 13 mm of air. Everything at high voltage is spaced out or potted. MIT's converter weighed about half a kilogram for 600 W, a record for that voltage at the time.

Can the battery be on board? Yes.

Battery mass

Wh stored

Can a solar panel be on board? Marginal.

Panel area on the wing Sunlight

W from the panel

The battery flew. Carrying the power supply was the whole point of the 2018 flight; the tethered hobby lifters of the 2000s had a cable to a supply on the ground. The 54 small cells weighed roughly a quarter kilogram and held about 30 Wh: enough for several minutes of full thrust, far more than the twelve seconds it took to cross the gym.

Solar is the tempting next step, and MIT's paper says as much. Lightweight cells at about 22 % efficiency give roughly 220 W per square metre in full noon sun, and the 5 m wing has about 2 m² of upper surface. That is close to what the plane needs, with no fuel or battery to run down, but only at noon, only in clear sky, and with the panel's own weight and a battery for launch still to carry. For the drone, needing ten times the power, a wing-sized panel is nowhere near enough.

6 More than 40 kV?

Table of these values

What goes up. Above onset the current grows roughly with the square of the voltage, and thrust follows it. Push the section 1 slider past 40 kV and each metre of wire makes more. Doubling the voltage gives nearly four times the thrust from the same wire.

What goes down. Thrust per watt is gap ÷ (mobility × voltage), so it halves when the voltage doubles. More thrust from the same wire costs disproportionately more battery, and the ionic wind's own energy is a smaller and smaller share of what the converter puts out.

The wall. Every gap has a sparkover voltage, roughly 1.2 kV per millimetre in dry, clean air and lower in humidity, dust or rain. Past it the corona turns into an arc: the air becomes a conductor, the converter's output collapses, thrust goes to zero, and the balsa and Kevlar are near a spark. Try 65 kV at the default 5 cm in section 1.

The way out is a wider gap. Both the sparkover and the thrust per watt scale with the gap, so the real design rule is: pick the field the air tolerates, about 0.8 kV/mm on the MIT plane (40 kV across 5 cm), then choose the widest gap and highest voltage the airframe can carry. Widen the gap slider and the wall on the chart moves right.

Why MIT stopped at 40 kV. The multiplier ladder happily grows more rungs; the ladder is not the problem. The problem is that every extra kilovolt needs more clearance on the airframe (13 mm of air per 40 kV, more through wet air), heavier potting in the converter, and more stray corona from any sharp edge, and each of those adds mass to a 2.5 kg aircraft. The published record is 40 kV at about 0.5 kg of converter. Higher voltages are used on the bench, with bigger gaps, where nobody has to lift the insulation.

7 Compress the air first?

Ram compression from speed

Airspeed

Pa of ram pressure

Compress it with a machine

Pressure rise

W to compress the stream

Or fly where the air is dense

Altitude

kV sparkover for this gap

What denser air does at the electrodes. Sparkover and corona onset both rise in proportion to density, so a denser gap takes more voltage before it arcs. Ion mobility falls in proportion, so at a fixed voltage the current drops, the thrust stays about the same, and thrust per watt improves by the same factor. Real gains, but modest: ten percent denser air is worth ten percent.

What it costs. An ion thruster is a pressure-rise device of a few pascals over a large area: the MIT plane made about 3 Pa. Ram compression at 4.8 m/s is about 14 Pa, a hundredth of a percent of an atmosphere, and only reaches a few percent near 100 m/s, where a slow ionic wind is useless anyway. A mechanical compressor is a fan, the thing the aircraft exists to avoid, and raising a few cubic metres per second by even one percent of an atmosphere costs kilowatts, ten times the whole plane's power. A sealed pressurised chamber does not work either: the thruster must exhaust to the atmosphere, so you would have built an electrostatic jet, throwing a little air fast, the opposite of the wide slow flow that makes ionic wind efficient.

The form of the idea that works. MIT's later thrusters put several emitter–collector stages in series inside a duct, each adding its few pascals to the same stream. That is the thruster raising its own pressure, not a stage ahead of it.

What the density argument really says is that ion planes want dense air: sea level on a cold day. At 10 km the air is a third as dense, this page's default gap arcs at about 20 kV instead of 60, and the efficiency drops with it. An ion aircraft is a low-altitude machine.

8 Notes

Solid-state propulsion means the thruster has no moving parts, the same sense as a solid-state drive. Ion engines on spacecraft use the same idea with their own propellant; in the atmosphere the air itself is the propellant, which is why it works at all here and not in space.
The model on this page is a one-dimensional textbook approximation: corona onset grows with gap, current follows the Townsend relation I ∝ V(V − V₀), and thrust is I·d/μ with the ion mobility μ = 2×10⁻⁴ m²/(V·s). It is calibrated so 20 m of wire at 40 kV across 5 cm gives 3.2 N for 512 W, the MIT plane's published figures (6.25 N/kW).
The airplane: H. Xu et al., "Flight of an aeroplane with solid-state propulsion", Nature 563, 532–535 (November 2018). Wikipedia summary. The high-voltage converter came from David Perreault's power-electronics group; MIT News describes it.
The drone: Undefined Technologies' Silent Ventus prototypes, reported by New Atlas (4.5-minute flight, 2022).
Older cousins: the tethered "lifter" hobby craft of the 2000s and the 1960s Brown-Biefeld experiments used the same corona-plus-collector idea, but always needed a power cable from the ground. Carrying the power supply on board is what the 2018 flight added.