The Propeller Effect

A spinning star that refuses to be fed

Picture a garden sprinkler, or mud flung off the tread of a fast-spinning bicycle wheel. Drop a clod onto the rim and it is not held; it is hurled tangentially outward. A rapidly rotating, strongly magnetised neutron star can do exactly this to the gas streaming toward it from a companion star. The infalling plasma never reaches the surface. Instead it is gripped by the star's whirling magnetic field, dragged up to a velocity faster than it can orbit, and slung back out. The result is a paradox: a compact object surrounded by a reservoir of gas, with a clear gravitational appetite, that nevertheless stays dim — because it is actively rejecting its food.

This is the propeller effect. It is not a small correction to accretion physics; it is a switch. The same neutron star, fed at a slightly higher rate, would blaze in X-rays as gas crashes onto its magnetic poles. Fed a little less, the same magnetic field that channelled the gas now becomes a barrier that throws it away. The boundary between "swallowing" and "flinging" is sharp, and it is set by a simple competition between two radii.

The two radii that decide everything

Two characteristic radii govern magnetised accretion, and their ratio decides whether the star eats or ejects.

The first is the magnetospheric radius (often called the Alfvén radius), r_m. This is where the magnetic field is strong enough to halt and redirect the inflow — where magnetic stress balances the ram pressure of the falling gas. For spherical accretion the standard estimate is

r_m ≈ ( μ⁴ / (2 G M Ṁ²) )^(1/7)

where μ = B R³ is the magnetic dipole moment, M the neutron-star mass, and Ṁ the accretion rate. Note the weak dependence: r_m ∝ Ṁ^(−2/7), so a hundredfold drop in accretion rate moves the magnetosphere out by only a factor of ~3.7. Inside r_m the gas is forced to follow field lines and corotate with the star; outside it the gas orbits freely.

The second is the corotation radius, r_c. This is the radius at which a circular Keplerian orbit has exactly the same angular velocity as the spinning star — the one orbit that matches the star's spin period P:

r_c = ( G M P² / 4π² )^(1/3)

Inside r_c, a corotating field line moves slower than a free orbit at that radius; matter locked to it lags the Keplerian speed and can settle inward. Outside r_c, a corotating field line moves faster than the local orbital speed; matter locked to it is being whipped beyond the speed needed to stay bound. The decisive comparison is therefore between r_m (where the field grabs the gas) and r_c (where corotation crosses the orbital speed):

r_m < r_c   →   ACCRETOR   (gas channelled to the poles, bright X-rays)
r_m > r_c   →   PROPELLER  (gas flung out, centrifugal barrier)

A convenient dimensionless form is the fastness parameter ω_s, the ratio of the stellar angular velocity to the Keplerian angular velocity at the magnetospheric radius:

ω_s = Ω_star / Ω_K(r_m) = (r_m / r_c)^(3/2)

The propeller switches on when ω_s > 1. A "slow rotator" (ω_s ≪ 1) accretes; a "fast rotator" (ω_s > 1) propels.

The energetics — work done by the field

The reason the gas leaves with more energy than it arrived with is that the magnetic field does positive work on it. At the magnetospheric radius the gas is forced onto field lines rotating at Ω_star. If r_m sits outside corotation, the field line at r_m is moving at azimuthal speed v_φ = Ω_star r_m, which exceeds the local Keplerian speed v_K = √(GM/r_m). The specific kinetic energy imparted by rigid corotation,

½ (Ω_star r_m)² = ½ Ω_star² r_m²

can exceed the local gravitational binding energy GM/r_m. When the corotation speed surpasses the local escape speed (v_esc = √(2GM/r_m)), the gas is formally unbound and leaves the system; this is the "strong propeller." When it exceeds Keplerian but not escape, the gas is pushed out to a larger, still-bound radius and may pile up there — a "weak propeller" or trapped-disk configuration. Either way, angular momentum flows from the star into the gas, so the propeller spins the neutron star down. The torque is roughly

N ≈ −Ṁ √(G M r_m)  ×  (something of order (ω_s − 1))

negative because the star loses angular momentum carried off with the ejected matter. Over time this drives the system toward spin equilibrium, where r_m ≈ r_c and the net torque vanishes. Setting r_m = r_c and solving gives the equilibrium spin period

P_eq ∝ μ^(6/7) Ṁ^(−3/7)

This is one of the most important results in compact-binary astrophysics: the long-term spin of an accreting pulsar is dictated by its magnetic moment and its accretion rate, not by how fast it was born.

The key numbers

The propeller effect operates across an enormous range of neutron-star magnetic fields and spin periods. Some representative values:

Quantity	Typical value	Notes
Neutron-star mass M	1.4 M☉ (≈2.8 × 10³³ g)	Canonical; radius R ≈ 10–12 km
Surface field B	10⁸ G (recycled MSP) – 10¹³ G (young X-ray pulsar)	Magnetars reach 10¹⁴–10¹⁵ G
Dipole moment μ = B R³	10²⁶ – 10³¹ G cm³	For R = 10⁶ cm
Spin period P	1.4 ms (fastest MSP) – seconds to minutes	Sets corotation radius
Corotation radius r_c	~25 km (P = 1.6 ms) to ~4 × 10⁴ km (P = 100 s)	r_c ∝ P^(2/3)
Magnetospheric radius r_m	~10⁷–10⁹ cm	r_m ∝ μ^(4/7) Ṁ^(−2/7)
Propeller luminosity threshold L_prop	~10³⁴–10³⁷ erg/s	Scales with B and P

For a canonical accreting X-ray pulsar with B = 10¹² G (μ ≈ 10³⁰ G cm³) and spin period P = 3 s, corotation sits at r_c ≈ 3.5 × 10⁸ cm, and the propeller turns on when the accretion rate falls below roughly 10¹⁵–10¹⁶ g/s — corresponding to an X-ray luminosity of a few × 10³⁴–10³⁵ erg/s. Below that, the source does not fade gradually toward zero; it drops abruptly, often by one to three orders of magnitude, as the magnetosphere swings outside corotation and the centrifugal gate slams shut.

Observing the propeller — the luminosity drop

The cleanest observational signature is a sharp, threshold-like collapse in X-ray brightness during the decay of an outburst. As an accreting neutron star transient fades, Ṁ drops, r_m expands as Ṁ^(−2/7), and at the moment r_m crosses r_c the source falls off a cliff. The luminosity at which this happens is set by the condition r_m(Ṁ) = r_c(P):

L_prop ≈ 4 × 10³⁷  ξ^(7/2)  B₁₂²  P_s^(−7/3)  (M/1.4 M☉)^(−2/3)  (R/10⁶ cm)⁵   erg/s

where B₁₂ is the surface field in units of 10¹² G, P_s the spin period in seconds, and ξ ≈ 0.5–1 a geometry factor relating the disk magnetospheric radius to the spherical Alfvén radius. This strong scaling — L_prop ∝ B² P^(−7/3) — means slow-spinning, strongly magnetised pulsars hit the propeller at high luminosity, while fast millisecond pulsars only propeller at very low rates. Measuring the threshold luminosity in several sources therefore yields an independent estimate of the neutron star's magnetic field, in good agreement with field strengths inferred from cyclotron lines.

Below the propeller threshold the source is not perfectly dark. A residual luminosity persists — from matter leaking through the barrier, from the cooling neutron-star surface, and from the small fraction that still reaches the poles. This "propeller floor" is itself a diagnostic: its level distinguishes a true centrifugal barrier from a simple drop in fuel supply.

Worked example: does SAX J1808.4-3658 enter the propeller?

Take the accreting millisecond X-ray pulsar SAX J1808.4-3658, discovered in 1998 — spin period P = 2.49 ms (frequency 401 Hz), mass M ≈ 1.4 M☉, surface field inferred B ≈ 1–2 × 10⁸ G (μ ≈ 1–2 × 10²⁶ G cm³). First, the corotation radius:

r_c = (G M P² / 4π²)^(1/3)
    = (6.67×10⁻⁸ × 2.8×10³³ × (2.49×10⁻³)² / 39.5)^(1/3)
    ≈ 3.1 × 10⁶ cm  ≈ 31 km

So the corotation radius is only about three neutron-star radii out — extremely close in, because the star spins 401 times a second. Now ask: at what accretion rate does the magnetospheric radius r_m, growing as Ṁ falls, reach this 31 km? Using r_m ≈ (μ⁴/2GMṀ²)^(1/7) with μ = 1.5 × 10²⁶ G cm³ and solving r_m = r_c for Ṁ gives Ṁ_prop of order 10¹⁵ g/s, i.e. an X-ray luminosity

L_prop ≈ G M Ṁ_prop / R ≈ (6.67×10⁻⁸ × 2.8×10³³ × 10¹⁵) / 10⁶
       ≈ 2 × 10³⁵ erg/s

And indeed, SAX J1808.4-3658 is observed to drop sharply out of its outbursts near this luminosity, with the decay steepening exactly where the propeller model predicts. The weak field (only ~10⁸ G) is what pulls the threshold luminosity down low and the corotation radius in close — a direct consequence of L_prop ∝ B² P^(−7/3): a recycled millisecond pulsar with a buried field only propellers at very faint levels, whereas at comparable spin a young 10¹² G pulsar — with a field some four orders of magnitude stronger — would gate at a threshold higher by the corresponding factor of B².

The accretor–propeller–ejector sequence

The propeller is the middle stage of a three-regime classification worked out by Roger Davies, Andrew Fabian and James Pringle in the late 1970s. As the accretion rate falls (or the star spins up), a magnetised neutron star moves through:

Regime	Condition	What happens to the gas	Observational state
Accretor	r_m < r_c	Channelled to magnetic poles, lands on surface	Bright X-ray pulsar / accreting MSP
Propeller	r_c < r_m < r_lc	Centrifugally flung out at the magnetosphere	Faint, gated; sharp luminosity drop
Ejector (radio pulsar)	r_m > r_lc	Pulsar wind sweeps it away; no infall reaches magnetosphere	Rotation-powered radio pulsar

Here r_lc = c P / 2π is the light-cylinder radius, where corotation would require the speed of light. When the magnetosphere expands beyond the light cylinder the star can no longer be confined by the inflow at all and turns on as a radio pulsar, its relativistic wind clearing the surroundings. The propeller is thus the gatekeeper between a quietly fed neutron star and a fully switched-on pulsar — and a system near spin equilibrium can flicker back and forth across these boundaries as its accretion rate fluctuates.

History, key objects, and missions

The mechanism was introduced by Igor Illarionov and Rashid Sunyaev in 1975, in the same fertile period that produced the Shakura–Sunyaev disk model. They recognised that for a sufficiently fast rotator the magnetosphere acts as a centrifugal barrier and named the resulting "propeller" expulsion of matter. Davies, Fabian & Pringle (1979) formalised the accretor/propeller/ejector regimes for isolated and binary neutron stars. The trapped-disk and "dead disk" refinements, in which matter piles up just outside corotation instead of being expelled outright, were developed by Spruit & Taam (1993) and revisited by D'Angelo & Spruit (2010).

Observationally, the propeller moved from theory to measurement with X-ray timing missions. RXTE (the Rossi X-ray Timing Explorer, 1995–2012) caught the abrupt outburst decays of accreting millisecond pulsars like SAX J1808.4-3658. The strongest modern evidence comes from transitional millisecond pulsars — systems caught switching between rotation-powered radio pulsar and accretion-powered states — the prototype being PSR J1023+0038, whose 2013 state change was tracked by Fermi-LAT, Chandra, XMM-Newton and ground-based radio telescopes. The eclipsing accretor 4U 0115+63 and the slow pulsar in GX 1+4 show propeller-driven luminosity gaps and torque reversals respectively. More recently NuSTAR and NICER have measured propeller thresholds in transient pulsars and in the brightest neutron-star ultraluminous X-ray sources, where a super-Eddington accreting neutron star with a strong field sits right at the propeller boundary.

Variants and related phenomena

• Strong vs. weak (trapped-disk) propeller. If corotation speed at r_m exceeds the escape speed, gas is unbound and ejected from the system (strong propeller). If it exceeds Keplerian but not escape, gas is pushed to a larger bound radius and accumulates — a trapped or "dead" disk that can episodically leak inward, producing low-level flickering.
• White-dwarf propeller. The same physics applies to magnetic cataclysmic variables. The intermediate polar AE Aquarii is the textbook "magnetic propeller": its white dwarf spins once every 33 seconds, fast enough that its corotation radius lies inside the magnetosphere, and it expels most of the transferred matter rather than accreting it.
• Spin-up / spin-down torque reversals. Near spin equilibrium the sign of the torque depends sensitively on ω_s. Sources like GX 1+4 and 4U 1626-67 have been observed to switch between secular spin-up and spin-down — the propeller torque flipping as the accretion rate drifts across the equilibrium value.
• Recycling and millisecond pulsars. The spin-equilibrium endpoint of long-lived, low-field accretion is a millisecond pulsar. A neutron star with a low recycled field (~10⁸ G) reaches very short equilibrium periods; once accretion stops, it emerges as a rotation-powered radio MSP. The propeller barrier is the physics that links accretion rate to final spin.
• Magnetar fallback propellers. In some models a young magnetar surrounded by supernova fallback enters a strong propeller phase that dumps enormous spin energy into the ejecta, a proposed power source for some superluminous supernovae and long gamma-ray bursts.

Common misconceptions and subtleties

• "The propeller blocks gas with magnetic pressure." Not quite — it is the rotation of the field that ejects the gas, not magnetic pressure alone. A slowly spinning star with the same field strength simply channels the gas to its poles. The propeller requires the corotation speed at r_m to beat the local orbital speed; it is a centrifugal, not a magnetostatic, barrier.
• "In the propeller state the source goes completely dark." It usually does not. A residual "propeller floor" luminosity persists from leakage through the barrier, the cooling surface, and matter that still trickles to the poles. The signature is a sharp drop, not a true zero.
• "The Alfvén radius and corotation radius are the same thing." They are independent quantities. The magnetospheric (Alfvén) radius depends on field and accretion rate (r_m ∝ μ^(4/7) Ṁ^(−2/7)); the corotation radius depends only on mass and spin (r_c ∝ M^(1/3) P^(2/3)). The propeller is precisely about which of these two is larger.
• "Disk accretion and spherical accretion give the same r_m." The disk magnetospheric radius is smaller than the spherical Alfvén radius by a factor ξ ≈ 0.5, because a disk delivers its ram pressure more efficiently. Real fits to propeller thresholds carry this ξ as an order-unity uncertainty.
• "The propeller always unbinds matter." Only the strong propeller does. Whether gas is fully ejected, parked in a trapped disk, or recycled depends on how far r_m sits outside corotation — i.e. on the magnitude of ω_s, not just its sign.