Magnetic Sail Propulsion

Speculative If a magnetic sail delivers in space the thrust its models predict — which nothing has ever measured, and 7 says what would settle it — the prize is a propulsion system whose collecting area is a field rather than a structure, with an effective areal density four to five orders of magnitude below a membrane's. Take the concepts' own arithmetic. A one-newton electric sail working against 2 nanopascals of solar-wind dynamic pressure has to intercept about 5 × 10⁸ square metres, or 500 square kilometres. The hardware doing it is some 20 kilograms of wire: 2,000 kilometres of aluminium at roughly 10 grams per kilometre. That is an effective areal density near 0.04 milligrams per square metre, and under a milligram even counting the whole 50–200 kilogram propulsion unit. Sail membranes are quoted in grams per square metre.

Frontier The second half of the prize is how the thrust falls off, and it is the half least often stated. Photon pressure drops as 1/r². Electric-sail thrust is modelled as 1/r, because the sheath around a charged wire widens as the plasma thins. Magnetic-sail thrust, taken from the pressure-balance relation the concept literature uses, falls as roughly r^−4/3 — that exponent is arithmetic on a stated relation, not a measured law. At 5 astronomical units those rules give 4%, 20% and about 12% of the 1 AU thrust. A drive that still pushes past Jupiter is a different vehicle from one that does not.

Frontier The gain is narrow and it accrues to one mission shape: fast, one-way, outbound. The published concept accelerations run 0.5–5 mm/s² for electric sails and 1–10 µm/s² for magnetic sails against 0.1–1 mm/s² for photon sails — a spread of three orders of magnitude inside one family, which is itself the honest headline. A European Space Agency study budgets 0.39 newtons and 1.0 mm/s² for a 391-kilogram asteroid tourer on 409 watts. Nobody gets a cargo route, a return trip or station-keeping: the thrust points away from the Sun and the vehicle ends up travelling with the wind. What a heliophysics or outer-planet programme would get is flyby cadence at a small-satellite price.

Handwave The condition is the whole brief: not one of those numbers has been measured against a real solar wind. Four electric-sail tether missions have flown and none deployed a tether. The only force measurements in this family were made in vacuum chambers at roughly 10⁵ below the scale the concepts need. The bottleneck list in 5 makes that concrete; the gap between a simulated 500 nanonewtons per metre and a measured one is what this brief is organised around.

Established A magnetic sail exchanges momentum with the solar wind plasma rather than with sunlight, and that one substitution changes what the vehicle is made of. The Sun sheds a supersonic, fully ionised flow — protons, electrons and alpha particles at 250–750 kilometres per second with 3–10 particles per cubic centimetre near Earth — whose dynamic pressure at 1 astronomical unit is typically 1–6 nanopascals. A spacecraft that deflects that flow takes its momentum. Because charged particles are deflected by fields and fields extend far beyond the conductor that makes them, the deflecting surface does not have to be built, folded, or held flat.

Established The governing relation is a pressure balance, and it is textbook. A spacecraft magnetic field stands the wind off at the radius where magnetic pressure equals the wind's dynamic pressure: B²(R_m)/2µ₀ ≈ ½ρv², and the force follows as roughly πR_m²ρv². For a dipole, B falls as 1/r³, so R_m scales as the cube root of the magnetic moment and the intercepted area as its two-thirds power.

Established Magnetic moment is bought with current and loop area, not with structural mass, which is the entire mechanical argument for the family.

Established Three variants share that momentum source and almost nothing else. The magsail, proposed by Dana Andrews and Robert Zubrin around 1990, is a single superconducting loop tens of kilometres across, carrying enough current to stand the wind off hundreds of kilometres out. Mini-magnetospheric plasma propulsion (M2P2) and its descendant the plasma magnet replace most of that loop with plasma: a small antenna drives currents in an injected or entrained plasma, and those currents carry the field outward, so the coil shrinks by three or four orders of magnitude while the field region does not.

Established The electric sail, invented by Pekka Janhunen in 2006 at the Kumpula Space Centre, abandons magnetism entirely: long, thin wires held at tens of kilovolts positive repel solar-wind protons electrostatically through a plasma sheath tens of metres wide.

Frontier The electric sail is in this brief because the momentum source is the same, not because the physics is. Its interaction length is the Debye sheath, which simulations put near 20 metres for a 10-micrometre wire in average 1 AU plasma, and its thrust is a Coulomb-drag effect on individual protons rather than a fluid deflection. It also carries most of the family's hardware experience, because it is the only variant anyone has tried to fly. Grouping the three is a choice this brief defends explicitly in 15: they fail for different reasons and should not be quoted as one technology.

Established What this brief does not own. Photon-pressure sailing belongs to Solar Sail Systems, and the family survey — ion, Hall, magnetoplasmadynamic, electrodynamic tether — belongs to Electrodynamic Propulsion Concepts, which places magnetic sails inside it. Using a magnetic field to shield rather than to push is Artificial Magnetospheres. This brief takes the one question those three defer: what is the measured thrust, and why after thirty-five years is there still none.

Established Start with the only quantity in this subject that is genuinely well measured: the wind itself. Merged upstream observations in the OMNI dataset, running from late 1963, give a mean bulk speed of 446.6 kilometres per second and a mean proton density of 6.92 per cubic centimetre at 1 astronomical unit; Toivanen and Janhunen used roughly 29 years of gap-free coverage to drive an electric-sail thrust model directly from observations rather than from an assumed average.

Established The Deep Space Climate Observatory has held the Sun–Earth L1 point since 8 June 2015 with PlasMag — a Faraday cup, an electrostatic analyser and a magnetometer — and the Advanced Composition Explorer did the same job before it. An upstream monitor with minute-cadence density, speed and field is not a thing this field needs to build; it has been running continuously for three decades.

Established The anchor number checks out, with one qualification worth stating precisely. At a solar irradiance near 1,370 watts per square metre the photon momentum flux is about 4.6 micropascals if the light is absorbed and 9.1 if it is perfectly reflected. Against a solar-wind dynamic pressure of 1.6 nanopascals those ratios are about 2,900 and 5,700; against 2 nanopascals, about 2,300 and 4,600.

Established So “the solar wind carries roughly one five-thousandth of the momentum flux of sunlight” is right for a reflecting sail and overstates the gap by a factor of two for an absorbing one, and the wind's own 1–6 nanopascal range moves the ratio by a factor of six on its own. The commission's anchor survives; the number is a band, not a constant.

Established The flight record is short enough to state completely: no magnetic sail, plasma magnet or electric sail has ever produced a measured thrust in space. Four electric-sail tether experiments have flown. ESTCube-1 (2013–2015) failed when the piezoelectric motor that was to turn the reel did not turn it. Aalto-1, launched June 2017, never deployed its tether and re-entered in 2024. FORESAIL-1, launched May 2022, lost contact after two weeks, before any electric-sail experiment could be performed; a rebuild, FORESAIL-1p, is under development.

Established ESTCube-2 launched from Kourou on a Vega vehicle on 9 October 2023, communication was never established, and Arianespace subsequently confirmed to the team that the satellite had failed to separate, most likely remaining attached to the upper stage until it burned up on re-entry. Four missions, zero tethers deployed, zero newtons measured.

Frontier Even a clean success in that programme would not have measured what the commission asks for, and this is the structural point the field states least clearly. ESTCube-2 carried 30 metres of tether to a roughly 565-kilometre orbit to demonstrate centrifugal deployment, charging against spin, the positive electric-sail mode with an electron emitter, and an ionospheric plasma brake — where a 300-metre tether was projected to shorten deorbiting time by an order of magnitude. But low Earth orbit has no solar wind. The interaction there is Coulomb drag against the co-rotating ionosphere, which shares the tether physics and none of the flow. A tether test in LEO measures a brake, not a sail, which is why the follow-on concepts — ESTCube-LuNa's 2-kilometre tether in lunar orbit, the ESME software and experiment design behind it — move the demonstration to where the wind actually blows.

Frontier The only force measurements in this family were made in vacuum chambers, and the best of them is worth quoting exactly. John Slough's NASA Innovative Advanced Concepts Phase II report on the plasma magnet states: the average thrust during the 200-microsecond pulse length was about 1.25 newtons, produced by the device interacting with a hydrogen plasma jet from a solar-wind simulator. Phase I had already driven over 10 kiloamperes of plasma current — up to 20 kiloamperes at higher rotating-field power — at an electron temperature near 18 electronvolts, with a 0.25-metre antenna at 108 kilohertz, and had shown the plasma structure standing off from the antenna with no significant contact.

Frontier It had also hit the wall, literally: expansion stopped at 0.2 metres because the conducting chamber wall prevented it, against a design target of 30 kilometres in space — a scaling gap of about 10⁵.

Frontier Slough's own Phase II report names the three things that remain undemonstrated, and no programme since has demonstrated them. In the report's words: it must be shown that the plasma magnet can achieve an equilibrium configuration in the presence of a larger scale solar wind; that the expected thrust is actually imparted to the antenna structure; and that the system scales as predicted. Those three sentences, written for a NASA advanced-concepts programme two decades ago, are still an accurate statement of the field's position in 2026.

Frontier The adjacent magnetoshell work is the one place where a chamber test produced a large, clean effect: David Kirtley's 2012 Phase I report records a 2-metre argon magnetoshell showing a 1,000-to-1 increase in aerodynamic drag over bare hardware, with a measured effective interception diameter of 1.6 metres in a 2.2-metre chamber. That is aerocapture against a planetary atmosphere, not solar-wind propulsion, and the report is explicit that its low-plasma-beta dipole is a different regime from the high-beta configurations M2P2 assumed.

Speculative Everything else in the magnetic-sail literature is a design study, and the design studies disagree with each other by three orders of magnitude on the one quantity that decides the case. Andrews and Zubrin's 1990 treatment takes a 100-kilometre loop with a near-term superconductor current density of 10¹⁰ amperes per square metre, arrives at a 500-tonne coil, and quotes 1,980 newtons of thrust. Zubrin and Martin's January 2000 final report to NASA's Institute for Advanced Concepts tabulates an operational magsail at a 20-kilometre radius, 1.60 × 10⁴ amperes total, a current density of 2.27 × 10⁹ amperes per square metre, a coil mass of 7.06 × 10³ kilograms and a self-acceleration of 3.185 × 10⁻³ metres per second squared — which, multiplied out, is about 22 newtons on the bare coil. A 200-metre demonstrator is tabulated at 1.03 × 10³ amperes and 25.7 kilograms.

Frontier Against that, the 2025 review literature places magnetic sails at 1–10 micrometres per second squared, three hundred to three thousand times below Zubrin's own figure for the same class of device. Neither side is measuring anything; they are making different assumptions about effective blocking area.

Frontier M2P2 is the cautionary case, and it is the most instructive thing in the subject. Robert Winglee's 2000 concept put a 0.1-metre coil and a 2.5-centimetre solenoid in a roughly 50-kilogram package, injected plasma at 0.5 kilograms per day, and claimed 1 newton of thrust — against about 3 × 10⁻¹¹ newtons for the bare magnetic configuration, a claimed gain of 4 × 10¹⁰.

Frontier Three independent analyses took it apart. Khazanov's group (2003) found the sail too small and the thrust correspondingly tiny, with the field falling off closer to 1/r² than the assumed 1/r. Toivanen and colleagues (2004) concluded the thrust estimate was over ten orders of magnitude optimistic, because solar-wind momentum went into the magnetotail and current leakage rather than into the spacecraft. Cattell and colleagues (2005) found the required sail size, mass and magnetic flux all underestimated and the assumed falloff physically impossible. The concept that produced the family's most quoted number is the concept the literature has most thoroughly rejected, and the rejection is twenty years old.

Frontier The electric sail's headline thrust has its own revision history, and the field quotes the endpoint without the history. Janhunen's 2004 technical note proposed a 30 × 30 kilometre charged mesh at 6 kilovolts giving 1.8 newtons on 250 kilograms — 7 millimetres per second squared. The 2007 particle-in-cell study with Sandroos, using 7.3 protons per cubic centimetre, 12-electronvolt electrons and 400 kilometres per second, returned (5 ± 1) × 10⁻⁸ newtons per metre at 15 kilovolts. The 2008 performance analysis with Mengali and Quarta used 50–100 nanonewtons per metre and 100 tethers of 10 kilometres for 0.1 newtons on a 100-kilogram craft.

Frontier Then in 2009 Janhunen argued that trapped shielding electrons, which had suppressed the simulated sheath, are removed in practice by orbit chaotisation against the three-dimensional potential of the spacecraft body, and raised the estimate roughly five-fold to about 500 nanonewtons per metre at 20 kilovolts. The 2010 figure everyone quotes — 100 tethers, 20 kilometres each, 1 newton, 100 kilograms, 1 millimetre per second squared, thrust scaling as 1/r — rests on that argument.

Frontier To Janhunen's credit the revision was published with its own caveat, and the caveat is still outstanding. The 2009 paper states that the theoretical results presented call for experimental verification, and that verification could come from a measurement of electrosphere size, thrust force or both in a space or laboratory experiment. His companion conference paper is blunter about why theory alone will not close it: given the range of scales involved, a brute-force simulation approach is not fruitful, so while theory is essential and simulations helpful, experimental studies are crucial. The person whose model produced the number says the number needs an experiment. Seventeen years later there has been none.

Established What has been built and measured is hardware, not thrust, and the hardware record is respectable. The European Union's FP7 ESAIL project, coordinated by the Finnish Meteorological Institute across nine partners in Finland, Germany, Sweden, Estonia and Italy, ran to 30 November 2013 and took the key components to technology readiness level 4–5, including automatic production of a one-kilometre tether.

Established A 2024 paper reports two 60-metre flight tethers of 50-micrometre aluminium wire built with 6,480 twist bonds at 3.75-centimetre spacing, about 1.5 centimetres wide, each wire holding 49 grams in tension and the assembled tether tolerating a 14-gram pull. The European Space Agency's 2019 Technology Research Programme study budgets a real mission from these parts: 44 main tethers of 15.3 kilometres, 20 kilovolts nominal against a 40-kilovolt hardware limit, 0.39 newtons at 1 astronomical unit, 1.0 millimetre per second squared, 391 kilograms all-up, 409 watts, and a micrometeoroid-severance probability of 0.24% over five years for four-wire tethers.

Frontier The deepest open question is whether the interaction region is even large enough for the fluid picture the thrust estimates assume, and a one-line calculation shows why it matters. A 400-kilometre-per-second proton in a 50-nanotesla field — the magnetopause field the plasma magnet's own design assumes — has a gyroradius of about 84 kilometres. The plasma magnet's stated barrier radius is 4–10 kilometres. The sail is then an order of magnitude smaller than the turning circle of the particles it is supposed to turn, which is precisely the objection raised against M2P2 and is not obviously answered by making the coil out of plasma.

Frontier Zubrin's magsail, with an effective blocking radius the report puts near 100 kilometres, sits at the boundary rather than clearly inside it.

Frontier Whether momentum transferred to the plasma actually reaches the antenna is a separate question and was the fatal one for M2P2. Toivanen's group located the loss precisely: momentum went into the magnetotail and into current leakage rather than into the structure. Slough's Phase II report keeps the same item open in its own words — that the expected thrust is actually imparted to the antenna structure. This is not a modelling nicety. A field-based sail has no mechanical load path except the field itself, so a configuration that deflects the wind beautifully and transfers the reaction somewhere else produces a spectacular plasma structure and no thrust.

Frontier For the electric sail the live question is the trapped-electron population, and it is worth a factor of five to ten on the whole concept. Simulations that retain trapped electrons compress the sheath and give tens of nanonewtons per metre; the argument that the spacecraft body chaotises those orbits until the electrons collide with the tethers gives hundreds. Janhunen's own conference figure spans roughly 0 to 900 nanonewtons per metre across 0 to 40 kilovolts under the zero-trapped-electron assumption.

Frontier The two branches are not a range around a central value; they are different physical claims about a collisionless sheath, and one measurement of electrosphere size would settle which is right.

Speculative Whether spin and rotation buy real thrust is an open modelling claim with no experiment behind it. A 2023 particle-tracing study of a rotating magnetic sail reports thrust gains of 7.88 times at zero incidence and 19.38 times at 90 degrees relative to a non-rotating coil, for magnetic moments of 10⁷ to 10⁹ ampere square metres and rotation up to 1,000 revolutions per minute, with absolute thrusts of 40.6 micronewtons, 0.512 millinewtons and 8.51 millinewtons at those three moments.

Speculative Note what those absolute numbers do to the rhetoric: even the largest is eight millinewtons, four orders of magnitude below the newton-scale figures the family's headline claims use, for a device with a 100-metre coil at 10⁴ amperes.

Frontier Thrust vectoring is the difference between a drive and a slingshot, and it is only partly solved on paper. A photon sail tilts a rigid surface. An electric sail must tilt a spinning cartwheel of kilometre-scale wires by modulating individual tether voltages, and the published control work — spin-plane control, thrust vectoring with a realistic sail shape, schemes for controlling spin rate by the sail effect itself — is entirely analytical. A magnetic sail's attitude problem is worse in one respect: the thrust vector is set by the orientation of a magnetic dipole relative to the flow, and turning a multi-tonne superconducting loop is a gyroscopic problem nobody has built hardware for.

Frontier How variable the thrust actually is under real wind turns out to be a genuinely encouraging result, and it is measured rather than assumed. Driving the electric-sail model with observed OMNI conditions, Toivanen and Janhunen found that variations in acceleration are statistically small relative to variations in density and wind speed, because a sail that widens its sheath as the plasma thins self-reefs. Mean thrust came out near 5 × 10⁻⁸ newtons per metre with an initial acceleration of 0.5 millimetres per second squared and a floor near 0.43–0.46 during low-wind periods.

Established A navigator can plan against that. It is also, as of 2026, the closest thing this field has to a validated prediction — and it validates the model's variability, not its magnitude.

Speculative Interstellar deceleration is the frontier question with the largest payoff and the weakest floor. Perakis and Hein's 2016 analysis decelerates a craft from 5% of light speed to 35 kilometres per second using a magnetic sail at high speed and an electric sail at low, with a 50-kilometre maximum coil radius, superconductor at 6,000 kilograms per cubic metre and current density up to 2 × 10¹⁰ amperes per square metre. The combined system takes 29 years against 35 for the electric sail alone and 40 for the magnetic sail alone, at 8,250 kilograms total. At 0.1c the deceleration distance runs to about 2.5 light years — more than half the distance to Alpha Centauri, which is a constraint on mission design rather than a detail.

Speculative Interstellar Probes records that arrival deceleration is treated across the trade literature as a property of the drive rather than a technology in its own right; this is the technology.

Established The binding bottleneck is not physics, it is a reel. Four flights, four deployment-side failures: a piezoelectric motor that would not turn on ESTCube-1, a sail that never unfurled on Aalto-1, a spacecraft lost before the experiment on FORESAIL-1, and a satellite that never left the upper stage on ESTCube-2. The 2024 flight-tether paper describes the state of the art as two 60-metre tethers with 6,480 twist bonds, made robust specifically because earlier failures in the tether deployment mechanics or electronics led to unsuccessful deployments. The field's demonstrated deployed length in space is zero metres.

Frontier For the magsail the bottleneck is a superconducting loop nobody has any basis for deploying. Zubrin's operational design needs 7.06 tonnes of conductor at 2.27 × 10⁹ amperes per square metre in a 20-kilometre-radius hoop; the 1990 interstellar treatment assumes 10¹⁰–10¹¹ amperes per square metre and 500 tonnes at 100 kilometres. Nothing in the space-hardware record deploys a kilometre-scale rigid or tensioned superconducting ring, holds it against hoop stress from its own current, and keeps it cold.

Frontier Coil current density is the one input where the ground has genuinely moved, which is why High Temperature Superconductors is a real dependency rather than a courtesy link — but tape performance on a bench is not a deployed hoop.

Frontier The plasma magnet trades the coil bottleneck for a power-and-scaling bottleneck, and the scaling claim is enormous. Phase I sustained the laboratory configuration at 50–100 kilowatts in a 1-metre chamber. The space projection inverts that: with a 100-metre-scale antenna and a rotating-field power of 1 kilowatt, a space-based plasma magnet could produce a thrust power of about 700 megawatts, on an antenna the report masses at 20 kilograms. That is a leverage claim of roughly 10⁶ between input power and wind power intercepted, resting on the plasma carrying current out to 30 kilometres in a medium that has never been simulated at that scale with the kinetic physics that matters.

Speculative The earlier 2004 presentation of the same concept used a 10-metre antenna at 20 kilowatts for a 20-kilometre bubble and 4 newtons in simulation — the same device, three different antenna-power pairs, across three documents.

Frontier Survivability of a kilometre-scale wire is a quantified but untested bottleneck. The ESA study puts micrometeoroid severance at 0.24% over five years for a four-wire redundant tether, which is a number derived from flux models rather than from a tether that has flown. The tethers are 25–50 micrometres of aluminium; atomic-oxygen degradation ruled out silver as a bonding metal and cost ruled out gold. A single severance on a spinning multi-tether sail is not a graceful degradation; it is an unbalanced rotor.

Frontier Ground test is structurally unavailable for the magnetic variants, which is why this subject has stalled where it has. Slough's Phase I expansion stopped at 0.2 metres against a conducting wall, a compression of about 10⁵ relative to the intended 30-kilometre sail; the Phase II remedy was a dielectric chamber, which removes the wall but not the four orders of magnitude. No terrestrial facility produces a collisionless 400-kilometre-per-second flow over a volume kilometres across.

Frontier For the electric sail the equivalent problem is the sheath: a 20-metre Debye-scale structure around a 10-micrometre wire cannot be reproduced in a tank, which is exactly why the thrust-per-metre figure has been settled by argument rather than measurement for seventeen years.

Frontier Measuring millinewtons on a spinning spacecraft is itself a bottleneck and is under-discussed. A 100-kilogram craft at 1 millimetre per second squared changes velocity by 86 metres per second a day — easily visible in radiometric tracking, if nothing else is pushing. But solar radiation pressure on the bus, outgassing, thruster leakage and anisotropic thermal emission all act in the same direction, anti-sunward.

Established The experiment therefore has to be differential — charge on, charge off — and that is a requirement on the payload design, not an afterthought.

Frontier The first dependency is a kinetic plasma model that is believed by people outside the group that wrote it. Every thrust number in this brief comes from a simulation, and the simulations disagree by orders of magnitude depending on assumptions about trapped electrons, field falloff exponents and whether the magnetohydrodynamic approximation holds below the ion gyroradius. The M2P2 episode is what that looks like when it goes wrong: three independent groups reanalysing one concept and finding the falloff assumption unphysical. What the field lacks is not computing power but a benchmark problem with an agreed answer.

Established The second dependency is already met and the field should say so more often: upstream solar-wind metrology. DSCOVR at L1 since 2015, ACE before it, and the OMNI merged record back to 1963 give density, speed and field at cadence fine enough to normalise a thrust measurement against the flux that produced it. This is why the decisive experiment in 7 is cheaper than it looks: the reference instrument is somebody else's operational asset.

Established Space Weather Engineering owns that monitoring infrastructure and the case for extending it.

Frontier The third is superconductor performance per kilogram at cryogenic temperatures achievable without a large cooling plant. Perakis and Hein assume 6,000 kilograms per cubic metre and up to 2 × 10¹⁰ amperes per square metre for their coil; Zubrin's design runs an order of magnitude below that and still needs seven tonnes. The relevant advance is not peak field on a short sample but current density in long, light, deployable conductor. High Temperature Superconductors carries that record; this brief only consumes it.

Established The fourth is tether manufacture, and it is the one dependency with a clean upward trajectory. Ultrasonic wire-to-wire bonding of micrometre-diameter aluminium wires was published in 2011; a one-kilometre tether produced automatically followed in 2013; the 2024 flight tethers use twist bonding at 6,480 bonds per 60 metres precisely because ultrasonic bonding demanded cleanliness of the bonding heads and accurate wire placement that flight hardware kept failing to deliver.

Frontier The production problem is being solved by iteration on the ground while the deployment problem keeps failing in orbit.

Speculative The fifth dependency is one the field rarely lists: a demand signal. Electric and magnetic sails are best at the mission profile with the fewest customers — fast, one-way, outbound, no rendezvous, no return. NASA's Heliopause Electrostatic Rapid Transit System study, led by Bruce Wiegmann at Marshall with Janhunen, Nobie Stone, Rob Hoyt and Robert Schunk under the 2015 NIAC Phase II solicitation, projected 100 astronomical units in under 10 years and the heliopause at 120–150 astronomical units in under 15. That is a heliophysics mission with one obvious buyer and no queue behind it.

Frontier The decisive experiment is a single in-space deployment of a charged tether or a plasma magnet that returns a measured thrust referenced to the upstream solar-wind flux recorded by ACE or DSCOVR. Nothing else in this subject moves the assessment as far. Every performance number in 3 — 1,980 newtons, 22 newtons, 1 newton, 500 nanonewtons per metre, 1–10 micrometres per second squared — is a model output, and the models disagree by three orders of magnitude. One number with an error bar on it would do more for this field than another decade of simulation.

Established The measurement is well posed, which is the good news. Thrust is inferred from the difference between charged and uncharged intervals: spacecraft acceleration from radiometric tracking or an onboard accelerometer, or spin-rate and spin-plane change on a cartwheel configuration, differenced across a duty cycle that modulates tether voltage or rotating-field power on a timescale short compared with the solar wind's own variation. The normalising quantity — ρv² upstream, minute by minute — comes free from an operational L1 monitor.

Established The experiment's output is a coefficient, not a thrust: newtons per metre of tether per nanopascal, or intercepted area per unit magnetic moment, which is the quantity that can be extrapolated.

Frontier The confounders are the design problem, and they all push the same way. Solar radiation pressure on the bus, anisotropic thermal re-radiation, outgassing and residual propellant all act anti-sunward, which is where the sail thrust points. This is the error budget that sank several historical anomalous-acceleration claims, and it is why the duty-cycled differential design matters more than absolute sensitivity. A demonstration that reports a net anti-sunward acceleration without modulating the charge has measured its own thermal design.

Established Where it has to happen is settled by physics and is the reason four flights have proved nothing. Low Earth orbit has no solar wind; a charged tether there measures ionospheric Coulomb drag, which is a useful deorbiting device and a different experiment. The demonstration has to sit outside the magnetosphere — beyond roughly 10 Earth radii on the dayside, or in lunar orbit where the wind is uninterrupted except in the lunar wake, or on an escape trajectory.

Frontier ESTCube-LuNa is the concept built around exactly this recognition, taking a 2-kilometre tether to lunar orbit specifically because that is where the flow exists.

Frontier The minimum viable version is smaller than the community usually proposes. The ESME design work quotes a force of 1.51 × 10⁻⁴ newtons from a 2-kilometre tether charged to 20 kilovolts in average solar wind near the Moon, on a 15-kilogram spacecraft — about 10 micrometres per second squared, or 0.87 metres per second per day.

Frontier That is measurable by Doppler tracking within days, and it does not require the full 100-tether cartwheel, the 40-kilovolt supply or the 400-watt electron gun of the reference design. A single tether, a modest emitter and an honest duty cycle would discriminate between 50 and 500 nanonewtons per metre, which is the live disagreement.

Speculative For the magnetic variants the equivalent minimum experiment is harder and nobody has specified it publicly at this level of detail. A plasma magnet demonstration needs a rotating-field antenna, kilowatt-class power, and a plasma inventory the Phase I report puts at 1.8 milligrams of hydrogen for a 100-kilometre bubble, with 84 kilojoules of magnetic energy at 30 kilometres and a 50-nanotesla magnetopause field. Its success criterion is different from the electric sail's: before thrust, it has to show that the configuration reaches equilibrium in a real solar wind at all — the first of Slough's three undemonstrated items — which is an in-situ field and particle measurement rather than a dynamics measurement.

Frontier Two ground experiments would still earn their cost, and one is explicitly recommended in the existing literature. The first is the dielectric-walled chamber Slough's Phase II report names, which removes the conducting boundary that stopped expansion at 0.2 metres; it cannot close the 10⁵ scaling gap but it can test whether expansion continues as predicted once the wall is gone.

Frontier The second is a sheath measurement: Janhunen's 2009 paper says verification could come from a measurement of electrosphere size, thrust force or both, in a space or laboratory experiment, and electrosphere size is the cheaper half. A biased wire in a flowing plasma with a probe array mapping the potential structure would test the trapped-electron argument directly, at a cost measured in hundreds of thousands rather than tens of millions.

Speculative The natural experiment that would change the assessment without any hardware is a reanalysis nobody has published. Several spacecraft have flown long conducting booms and charged surfaces through the solar wind at L1 and beyond, and their tracking residuals are archived. An upper limit on unexplained anti-sunward acceleration from an existing mission, differenced against DSCOVR flux, would bound the Coulomb-drag coefficient from the bottom even if it measured nothing. The field has not done this, and it is the cheapest available result.

Handwave What no experiment on anyone's list would settle is whether the family scales to the sizes the headline claims need. A 2-kilometre tether measuring 150 micronewtons validates a coefficient; it says nothing about whether 2,000 kilometres of tether deploys, stays balanced, survives micrometeoroids for five years and holds 20 kilovolts. Nor would a plasma-magnet equilibrium at 30 kilometres tell you anything reliable about 100. Honesty requires separating the two: the decisive experiment settles the physics, and the engineering programme after it is where the newtons actually are.

Frontier The reference electric-sail vehicle is well specified on paper and small. The ESA Technology Research Programme study budgets 391 kilograms all-up, of which 143 kilograms is the sail including a 20% margin, driving 44 tethers of 15.3 kilometres at 20 kilovolts nominal for 0.39 newtons and 1.0 millimetre per second squared at 1 astronomical unit, on 409 watts.

Frontier Alternative statements of the same architecture give 100 tethers of 20 kilometres for 1 newton on 100 kilograms with a 400-watt electron gun, or roughly 2,000 kilometres of tether per newton with a propulsion unit of 50–200 kilograms. These are consistent to within a factor of two, which is better than anything on the magnetic side.

Frontier Spin is the structural system, and that is the design's cleverest and most fragile feature. There is no boom: the tethers are held out by rotation, which means deployment is a centrifugal process, attitude is gyroscopic, and thrust vectoring happens by modulating individual tether potentials around the spin. The ESA study's recommendations — bidirectional reeling for main tethers, full redundancy for auxiliary tethers — are aimed squarely at the failure mode the flight record keeps producing.

Established Compare the engineering posture with a photon sail's: Solar Sail Systems records booms that buckle and membranes that go pyramidal; the plasma family has no booms and no membranes, and has instead never got anything out of the canister.

Frontier The power system is modest and the voltage is not. Four hundred to 540 watts for electron emission is a small solar array at 1 astronomical unit; holding 20–40 kilovolts on exposed wire in a plasma, through eclipse and through coronal mass ejections, is a high-voltage engineering problem with little flight heritage. The Wind Rider concept's power profile shows the other end of the same issue: 1,300 watts at 1 astronomical unit falling to about 50 watts at Jupiter, on a 16U CubeSat chassis. A drive whose whole selling point is performance in the outer system has to work on the power available there.

Speculative The magnetic-sail vehicle has never been through an equivalent exercise, and the numbers that exist are coil tables rather than spacecraft. Zubrin and Martin tabulate a 200-metre, 25.7-kilogram demonstrator at 1.03 × 10³ amperes giving 3.25 × 10⁻⁶ tesla, and an operational 20-kilometre, 7.06-tonne coil at 1.60 × 10⁴ amperes giving 5.01 × 10⁻⁷ tesla.

Speculative There is no cryogenic system, no deployment mechanism, no attitude control and no structure in those tables, and the report's own summary is modest about it: with a small reduction in mass density from current values, such a wire would allow a magsail, without payload or support structure, to compete with solar sails. That sentence is doing a great deal of work.

Frontier The plasma magnet's engineering claim is the most attractive and the least substantiated. A rotating-field antenna of 100-metre scale massing 20 kilograms on 1 kilowatt, or 10 kilograms at laboratory scale, driving a 30-kilometre magnetosphere, is a mass budget any smallsat programme would take. The gap between that and the 50–100 kilowatts Phase I actually needed in a 1-metre chamber is the engineering question, and it is not a matter of efficiency: it is the claim that the plasma does the work of the coil once the scale is large enough.

Established The nearest neighbour is the one this brief is defined against. Solar Sail Systems owns photon pressure, and the contrast is instructive in both directions: photon sails have flown five times and have a measured, disappointing characteristic acceleration; magnetic sails have a large predicted one and have never flown. The two failure modes are opposite — one is a structures-and-control problem with real telemetry, the other is a physics-validation problem with none.

Established Electrodynamic tethers share the hardware and not the environment. Electrodynamic Propulsion Concepts surveys the family and places magnetic sails inside it; the ionospheric plasma brake that ESTCube-2 and Aalto-1 carried is the same wire doing a different job against a different plasma. The transferable asset is deployment and high-voltage experience, which is exactly what the flight record has failed to accumulate.

Frontier Magnetoshell aerocapture is the adjacent concept with the best laboratory evidence. Kirtley's NIAC work uses a magnetised plasma to raise drag against a planetary atmosphere — 1,000-to-1 over bare hardware in a 2-metre argon test, sized at 9 metres and 250 gauss for a Neptune capture producing 150 newtons on a 207-kilogram system at about 10 kilowatts peak.

Established It is deliberately low-plasma-beta where M2P2 was high-beta, and the report says so, which is why its numbers are not transferable to solar-wind sailing even though the hardware looks similar.

Established Shielding and propulsion use the same field for opposite purposes. Artificial Magnetospheres records what a deflecting magnetic structure costs when the requirement is radiation protection rather than thrust, and the mass and power figures there are a useful sanity check on any claim that a large magnetosphere is cheap.

Frontier Two further adjacencies shape the mission case rather than the physics. Beam-Powered Propulsion is the competitor for fast outbound missions and has the same structural advantage — leave the power source at home — with a different bottleneck. Deep Space Communications owns the link budget for anything that leaves at 300 kilometres per second and keeps going, which is the unglamorous constraint on every fast-flyby concept in this brief.

Established The institutional history of this subject is a NASA advanced-concepts programme funding it twice and stopping. The Institute for Advanced Concepts funded Zubrin's magsail study to a January 2000 final report, Slough's plasma magnet through Phase I and Phase II, and Kirtley's magnetoshell through Phase I in 2012; the Heliopause Electrostatic Rapid Transit System took a Phase II award in the 2015 solicitation.

Established Every one of those produced a final report and none produced a flight. That is what an advanced-concepts programme is for, but it also means the subject has been resourced at study level for a quarter of a century with no mechanism to convert a study into a demonstration.

Established Europe funded components rather than concepts, and got further. The FP7 ESAIL project, coordinated by the Finnish Meteorological Institute with nine partners across five countries, ran to 30 November 2013 and lifted key components to technology readiness level 4–5 with a one-kilometre tether produced automatically. ESA followed with a CleanSat building-block study on the electrostatic plasma brake and the 2019 Technology Research Programme asteroid-touring study. Component-level funding is why the electric sail has hardware and the magnetic sail has spreadsheets.

Frontier The flight attempts were funded as student and small-national missions, and that is both why they happened and why they failed. ESTCube-1, ESTCube-2, Aalto-1 and FORESAIL-1 were built by university teams in Estonia and Finland at CubeSat budgets. The resulting record — a motor that would not turn, a sail that did not deploy, a spacecraft lost in two weeks, a satellite that never separated — is a record of integration and operations margins, not of propulsion physics.

Frontier A propulsion-physics result requires a mission whose only job is the measurement, and no agency has bought one.

Speculative The private route exists and is under-capitalised by its own account. The Wind Rider team around Brent Freeze and Jeff Greason has taken the plasma magnet to conference abstracts — an August 2021 AIAA paper, a JOVE mission presentation in October 2021, an AGU Fall Meeting poster in December 2021 — and estimates development at tens of millions rather than hundreds of millions of dollars, with no NASA funding and a stated route through the decadal survey process. A 2021 projection of a late-2023 launch with full funding has not happened. Treat transit-time claims from this source as the proponent's own figures.

Frontier What the field actually needs institutionally is a metrology payload, not a mission. The decisive experiment in 7 is a rideshare-class instrument on any spacecraft leaving the magnetosphere: a tether, an emitter, a duty cycle and precise tracking. No agency has a line item for validating a propulsion coefficient on someone else's trajectory, and that gap is institutional rather than technical, which is why it appears as a requirement in 14.

Frontier The concrete near-term harm in this subject is orbital debris, and it is created by the test programme rather than by the technology. The electric-sail tether is 25–50 micrometres of aluminium in lengths measured in kilometres; a severed tether in low Earth orbit is a long, light, hard-to-track object in the regime where every operator now has a conjunction budget. The same hardware also mitigates the problem it creates: an ionospheric plasma brake of 300 metres was projected to shorten deorbiting time by an order of magnitude, which is the only application in this brief with a near-term commercial rationale.

Speculative Fast flybys and planetary protection sit in tension in a way this family sharpens. A drive that trades cheaply for speed and cannot brake produces high-velocity impactors on whatever it passes. The Wind Rider figures describe Jupiter in about three weeks and Neptune in 18 at asymptotic speeds near the wind's own 400 kilometres per second.

Speculative A spacecraft that cannot slow down is a kinetic penetrator by default, and the trajectory design rather than the drive is where that has to be handled.

Frontier The labour question is real and specific to how this field has been funded. Four flight failures were absorbed by student teams, and the cost of each was several years of graduate careers against a mission that returned no science. The published lessons-learned culture around ESTCube is genuinely good — in-orbit experience papers, honest failure attribution to deployment mechanics and electronics — and it is worth saying that the record is transparent even where the outcomes were poor.

Frontier The claims problem is the ethical issue most likely to damage the field. Newton-level thrust figures from M2P2 circulated for years after three independent analyses had found them wrong by up to ten orders of magnitude. Characteristic accelerations differing by a factor of a thousand appear in adjacent paragraphs of the same review literature. Quoting 1 newton or 500 nanonewtons per metre without the word “simulated” is not a rounding error; it is the difference between a research programme and a promise, and this brief flags it every time it appears.

Speculative If the coefficient is near the high end, the change is to outer-system cadence rather than to capability. The existing comparison points are blunt: New Horizons left Earth at about 45 kilometres per second and had slowed to 14 by the outer system; Juno peaked near 73. Concept studies for this family project the heliopause at 120–150 astronomical units in under 15 years against Voyager's 27 to 30, and asymptotic speeds approaching the wind's own. The civilizational content of that is not new destinations but many more visits to known ones, at smallsat prices, by teams that could not previously afford a flagship.

Frontier The asymmetry between going out and coming back is permanent and shapes everything downstream. The wind blows one way. A magnetic or electric sail can tack — the thrust vector tilts against the flow — but the whole family's advantage is anti-sunward, and the vehicle asymptotes to the wind speed rather than exceeding it.

Frontier This is an exploration technology, not a transport technology, and every economic argument built on it should be read with that constraint attached.

Speculative The interstellar case is the only one where this family has something no competitor offers. Deceleration at a destination is the unsolved half of every fast interstellar concept, and a magnetic sail is one of very few schemes that brakes against the destination medium rather than carrying propellant for it: 29 years from 5% of light speed using a magnetic sail at high velocity and an electric sail at low, on 8,250 kilograms. Against 0.1c the braking distance runs to roughly 2.5 light years, which does not so much solve the arrival problem as convert it into a constraint on cruise speed.

Handwave The counterfactual deserves equal weight, because it is currently the more likely one. Thirty-five years after the magsail paper, twenty after the first plasma-magnet chamber experiments, and seventeen after the thrust coefficient was revised upward by argument, the measured record is one 200-microsecond laboratory pulse. A field can sustain an internally consistent literature indefinitely without a measurement, and the honest civilizational implication of this subject in 2026 is that it is a live demonstration of how long that can go on.

These horizons track one thing above all — whether any device in this family has returned a thrust measured against upstream solar-wind flux — and then what follows if it has:

• 10 yr: Frontier Expect at least one more electric-sail tether deployment attempt, most plausibly the FORESAIL-1 rebuild or an ESTCube successor, and expect it to be in low Earth orbit measuring ionospheric Coulomb drag rather than solar wind. Speculative A lunar-orbit or escape-trajectory demonstration on the ESTCube-LuNa pattern — a 2-kilometre tether, roughly 150 micronewtons, a 15-kilogram bus — is the plausible first real measurement, and it is unfunded. Frontier Expect no magnetic-sail or plasma-magnet flight hardware in this window on current funding.
• 25 yr: Speculative If the coefficient is measured and lands near the high branch, expect electric-sail propulsion modules in the 0.1–0.5 newton class on outer-system flyby missions, on the architecture the ESA study already costs at 391 kilograms and 409 watts. Speculative If it lands near the 2007 low branch, expect the concept to be retained only for deorbiting, where the Solar Sail Systems drag-sail market already exists. Handwave A plasma-magnet equilibrium demonstrated in a real solar wind belongs in this window only if someone funds the dielectric-chamber work first.
• 50 yr: Speculative A superconducting magsail of the scale Zubrin tabulated — 20 kilometres, 7 tonnes — requires deployment technology that does not exist and a coil current density that High Temperature Superconductors may or may not deliver in light, long conductor. Handwave Regular fast outer-system access at the cadence the concept literature describes assumes both the physics and a demand signal that no agency has yet articulated.
• 100 / 250+ yr: Handwave Interstellar deceleration by magnetic sail is the horizon where this family is genuinely without competitors, and it rests on braking distances of light years, coils of tens of kilometres and superconductors nobody has made. Handwave The honest statement is that the 100-year projection is contingent on a measurement that could be made in the next five years for a few tens of millions of dollars, and has not been.

• Depends on The hard dependency on this map is High Temperature Superconductors, and it binds only the magsail branch: the coil tables in the concept literature assume current densities from 2.27 × 10⁹ to 2 × 10¹⁰ amperes per square metre in conductor light enough to deploy over tens of kilometres, which is a materials result rather than a propulsion one. Electrodynamic Propulsion Concepts carries the family survey this brief sits inside, and Space Weather Engineering owns the upstream monitoring — DSCOVR, ACE, the OMNI record — without which a thrust measurement cannot be normalised. The electric-sail branch waits on nothing scientific on this map: its dependencies are deployment mechanics and one uncontested measurement, both recorded below.
• Requires (not on this map) Six things, none of which is a brief. First, the number this entire subject turns on: newtons per metre of tether per nanopascal of upstream dynamic pressure, measured in a real solar wind and referenced to an L1 monitor — the live disagreement runs from 50 to 500 nanonewtons per metre at comparable voltages and has been settled by argument since 2009. Second, a plasma model trusted outside the group that wrote it for interaction regions smaller than the ion gyroradius: a 400-kilometre-per-second proton in a 50-nanotesla field turns on an 84-kilometre circle, and plasma-magnet designs assume barriers of 4–10 kilometres, which is the regime where the M2P2 thrust estimate failed by up to ten orders of magnitude. Third, a deployment mechanism that has actually worked in orbit: four flights have produced four non-deployments, and the demonstrated deployed length in space is zero. Fourth, tether supply at mission scale — a reference sail needs roughly 2,000 kilometres of bonded 25–50 micrometre aluminium against a demonstrated production record of one kilometre automatically and 60 metres in flight build. Fifth, an institutional home for a measurement rather than a mission: no agency funds a payload whose only product is a propulsion coefficient on someone else's trajectory. Sixth, a customer, because the profile this family is best at — fast, one-way, outbound, no rendezvous — is the profile with the fewest buyers.
• Enables No typed enabling edge is claimed, and the reason is the whole brief. A technology with no measured thrust in space cannot be recorded as enabling anything without converting an intention into a dependency. The candidate edges, if the coefficient lands high, are heliophysics access beyond 100 astronomical units and destination deceleration for Interstellar Probes, which records arrival braking as a gap the trade literature treats as a property of the drive. Both remain candidates.
• Adjacent Solar Sail Systems is the sibling with the opposite problem — flown hardware, measured disappointment. Artificial Magnetospheres uses the same physics for shielding and prices it. Beam-Powered Propulsion competes for the same fast-outbound missions from the same premise of leaving the power source at home, and Deep Space Communications owns the link budget every fast-flyby concept here quietly assumes. Outside this map: ionospheric plasma-brake deorbiting, magnetoshell aerocapture, high-voltage design in plasma, and the small-satellite industry that funded every attempt so far.

Speculative “Magnetic sails have demonstrated performance.” They have not. The entire measured record of the magnetic branch is laboratory: an average thrust of about 1.25 newtons during a 200-microsecond pulse against a hydrogen plasma jet in a vacuum chamber, with the device's expansion limited to 0.2 metres by a conducting wall, at a scale roughly 10⁵ below the design intent. No magnetic sail, plasma magnet or electric sail has produced a measured thrust in space, and the sentence is exact rather than rhetorical.

Frontier “M2P2 gives a newton of thrust from fifty kilograms.” This is the most-quoted number in the subject and three independent analyses rejected it between 2003 and 2005: Khazanov on sail size and a field falloff nearer 1/r² than 1/r, Toivanen and colleagues on momentum going into the magnetotail and current leakage with the estimate over ten orders of magnitude optimistic, and Cattell and colleagues on required size, mass and flux all underestimated with the assumed falloff physically impossible. The claim is not merely unverified; it has been specifically refuted, and it still circulates.

Established “The solar wind is a better push than sunlight because the particles have mass.” Per unit area it is worse by three to four orders of magnitude. Photon momentum flux at 1 astronomical unit is about 4.6 micropascals absorbed and 9.1 reflected; solar-wind dynamic pressure is 1–6 nanopascals.

Established The case for this family has never been pressure; it is that the collecting area is bought with magnetic moment or voltage rather than with membrane, so the areal density can fall far below anything a physical sheet allows. Confusing the two is the single most common error in popular accounts.

Frontier “A magsail needs no power.” A persistent-current superconducting loop needs no power to maintain its field and a great deal of everything else: charging, cryogenic maintenance, structure to react hoop stress, and attitude authority against a dipole torque. The plasma magnet explicitly needs continuous power — kilowatts in the optimistic projections, 50–100 kilowatts in the only laboratory configuration that has run. Only the electric sail is genuinely low-power, at 400–540 watts for electron emission, and it needs that power continuously or the thrust stops.

Speculative “Electric-sail thrust is 500 nanonewtons per metre.” That figure is a 2009 simulation result which replaced a 2007 simulation result of (5 ± 1) × 10⁻⁸ newtons per metre from the same author, on the argument that trapped shielding electrons are removed by orbit chaotisation against the spacecraft's potential structure. It may well be right. It has never been measured, the author said in the same paper that it calls for experimental verification, and the factor of five to ten between the branches is the difference between a useful drive and a curiosity.

Frontier “ESTCube proved the electric sail works.” ESTCube-1's reel motor failed in 2013; Aalto-1 never deployed and re-entered in 2024; FORESAIL-1 was lost two weeks after its May 2022 launch before any electric-sail experiment; ESTCube-2 never separated from the Vega upper stage after its 9 October 2023 launch and burned up attached to it. What the programme proved is that tethers can be manufactured — a kilometre automatically by 2013, two 60-metre flight tethers with 6,480 twist bonds by 2024 — which is a real result and a different one.

Speculative “A plasma magnet reaches 400 kilometres per second.” That is the solar wind's own speed, which is the asymptote a perfect drag device approaches and cannot exceed while sailing downwind. It is quoted alongside transit times — Jupiter in about three weeks, Neptune in 18, 550 astronomical units in under seven years — that are the proponents' own figures from a concept with no flight hardware and no agency funding. The comparison points are real: New Horizons left at 45 kilometres per second and slowed to 14, and Juno peaked near 73.

Handwave “It is propellantless, so it is close to reactionless.” It is not, and the distinction matters because the family is sometimes filed next to claims that do violate momentum conservation. A magnetic sail pushes on the solar wind and the solar wind pushes back; the momentum ledger closes against a medium that is measured minute by minute from L1. The interesting physics is entirely in how much area a field intercepts and whether the reaction reaches the structure — both ordinary, both hard, neither exotic.

Frontier “Rotating the coil multiplies the thrust by twenty.” A 2023 particle-tracing study reports gains of 7.88 and 19.38 times over a non-rotating sail at zero and 90 degrees incidence. The absolute numbers in the same study are 40.6 micronewtons, 0.512 millinewtons and 8.51 millinewtons for magnetic moments of 10⁷, 10⁸ and 10⁹ ampere square metres. A twenty-fold gain on a micronewton is still a micronewton-class device, and the result is a simulation of a configuration nobody has built.

Frontier “The concept literature agrees on performance.” It does not, and the disagreement is the most important fact in this brief. For the same class of device the published characteristic accelerations span 1–10 micrometres per second squared in the 2025 review literature against 3.185 × 10⁻³ metres per second squared in Zubrin's own NIAC tables — a factor of three hundred to three thousand — while electric sails are quoted at 0.5–5 millimetres per second squared. These are not error bars on a measurement; they are different assumptions about intercepted area, published without a measurement to arbitrate between them.