YORP Spin-Up: How Sunlight Tears an Asteroid Apart

Photons Push Back: The Mechanism, Step by Step

A photon of energy E carries momentum E/c. At 1 au the solar flux is about 1361 W m−², so the momentum flux of sunlight is only Φ/c ≈ 4.5 × 10−⁶ pascals. Nothing on Earth notices that. An airless rock with nothing else acting on it for ten million years notices it very much.

• Absorption. A facet tilted at θ to the Sun takes in (1−A)Φcosθ watts per square metre; Bennu's albedo A is just 0.044.
• Heating. The regolith warms to a few hundred kelvin — OSIRIS-REx measured Bennu's dayside between roughly 250 K and 400 K.
• Re-emission. Thermal infrared leaves a Lambertian surface beamed about the local normal, so the recoil is (2/3)εσT⁴/c per unit area, directed into the surface.
• Geometry does the rest. Incoming photons all arrive from one direction, so their torque about the spin axis largely averages away over a rotation. Outgoing photons leave along local normals, which on a lumpy body point in a systematically biased set of directions — a slight windmill bias in the facets leaves a residual twist.

That residual is YORP. It has a spin component that changes ω and an obliquity component that walks the spin axis toward asymptotic states near 0°, 90° or 180°. In the standard Rubincam approximation — zero thermal inertia, instantaneous re-emission — the spin component depends almost entirely on shape. Čapek and Vokrouhlický (2004) showed it is nearly independent of thermal conductivity, so spin-up predictions survive our ignorance of asteroid thermal properties while obliquity predictions do not.

Vladimir Radzievskii (1954) proposed an albedo-driven version; Stephen Paddack (1969) and John O'Keefe the shape-driven one that dominates for real asteroids; David Rubincam assembled them and coined the acronym in 2000. A third variant, tangential YORP (Golubov & Krugly 2012), comes from sunlight absorbed on a boulder's morning side and conducted sideways through the rock, so the evening-facing flank radiates harder and the recoil pushes along the direction of surface motion — which is why it almost always accelerates rotation.

The Governing Relation: Why Size and Distance Decide Everything

Write the torque as T ≈ CₕΦ(a)R³/c, where R is mean radius, Φ(a) the solar flux at heliocentric distance a, and Cₕ a dimensionless efficiency set purely by shape. For bulk density ρ the moment of inertia is I ≈ (8/15)πρR⁵. Divide, and everything follows:

dω/dt ∝ 1 / (ρ a² D²)

Double the diameter and the spin-up rate falls by four; move out to 2.5 au and it falls by another six. The torque grows as R³ while the inertia it must overcome grows as R⁵, and the asteroid loses.

Put numbers on it. (54509) YORP, the ~114 m object formerly called 2000 PH5, spins once every 12.17 minutes and accelerates at 2.0 × 10−⁴ deg day−². With a moment of inertia near 2 × 10¹² kg m², the torque is roughly 10−³ newton-metres — a gram-weight on a ten-centimetre lever. The sunlight its cross-section intercepts carries about 0.05 N of momentum flux, which at a 57 m lever arm could supply 3 N m, so YORP runs at an efficiency of a few parts in ten thousand. The uncancelled remainder is the whole story.

The time to reach the spin barrier scales as ρD²a². Bennu needs a few hundred thousand years; a 5 km main-belt asteroid at 2.5 au needs ~10⁹ years; a 10 km one runs to billions of years, comparable to the age of the Solar System. Hence YORP is a small-body phenomenon — asteroids above ~40 km still show the Maxwellian spin distribution expected from pure collisional evolution, while below it the distribution grows a marked excess of both very fast and very slow rotators.

The 2.2-Hour Wall and What Happens at It

A strengthless sphere held together only by gravity flies apart when equatorial centrifugal acceleration matches surface gravity: ω²R = GM/R². Substituting M = (4/3)πρR³ gives a critical period that depends on density alone, not size: P_crit = √(3π/Gρ). For ρ = 2.2 g cm−³ that is 2.23 hours — the spin barrier identified by Petr Pravec and Alan Harris in 2000. Plot period against diameter for thousands of asteroids and everything above ~11 revolutions per day is nearly empty for bodies larger than ~200 m. Below 200 m the barrier vanishes and objects rotate absurdly fast (1998 KY26 in 10.7 minutes, 2014 RC in 15.8 seconds), because a small enough body can be a coherent monolith.

The barrier is a statistical wall, not a law. (455213) 2001 OE84 is ~700 m across yet turns in 29 minutes, and (29075) 1950 DA, 1.3 km wide, turns in 2.12 hours — Rozitis, MacLennan and Emery (2014) showed its ~1.7 g cm−³ density demands a cohesive strength near 64 pascals, comfortably supplied by van der Waals forces between fine grains.

Because the limit tracks density, low-density rubble piles fail early: Bennu's 1190 kg m−³ (about 50% porosity) puts its critical period at 3.03 hours, not 2.2. Its present 4.296-hour spin already cancels roughly half the gravity at its mean radius — ~8 × 10−⁵ m s−² of attraction against a centrifugal term of ~4 × 10−⁵ m s−² — and appreciably more than half out on the equatorial bulge, with a strollable 0.20 m s−¹ escape speed.

Failure is not a bomb going off; it is a landslide in slow motion. Material creeps toward the equator, the body relaxes into an oblate top shape with an equatorial ridge, and grains that finally exceed escape leave at centimetres per second — slow enough that much debris re-accretes and some enters orbit. Models by Hirabayashi and Scheeres show the mode depends on where the strength is: a weak interior deforms, a weak surface avalanches. YORP also runs the other way, spinning bodies down until principal-axis rotation goes unstable and they tumble — the ~340 m (99942) Apophis is a plausible case, while classic large tumblers such as the ~4.5 km (4179) Toutatis are far too big for YORP and more likely owe their state to collisions or planetary encounters.

Measuring a Torque of One Millinewton-Metre

YORP is measured, not merely inferred. A constant angular acceleration makes rotational phase drift quadratically: Δφ = ½(dω/dt)t². Predict when each brightness maximum should fall across many apparitions and, if the body is accelerating, the maxima arrive progressively early, the offset growing as the square of the baseline.

For (54509) YORP that offset reaches about 200 degrees — more than half a rotation — in four years. Two teams reported it in the same 2007 issue of Science: Stephen Lowry's from optical photometry, Patrick Taylor's from Arecibo and Goldstone radar. The period shortens by about a millisecond per year, doubling the spin rate in roughly 600,000 years.

For Bennu the signal is far weaker: 3.63 ± 0.52 × 10−⁶ deg day−², accumulating only ~35 degrees of phase over the 1999–2011 baseline. Michael Nolan's team published it in 2019 and Carl Hergenrother's confirmed it with OSIRIS-REx OCAMS imaging from orbit: Bennu's 4.296-hour day is shortening by about a second per century, and it should reach its own critical spin in of order a million years.

The toolkit is multi-apparition photometry feeding the lightcurve inversion Mikko Kaasalainen developed; delay-Doppler radar from Arecibo (lost on 1 December 2020) and Goldstone DSS-14; thermal infrared from Spitzer and NEOWISE to pin sizes and albedos; and, for a few targets, spacecraft. About a dozen detections now exist, including (1862) Apollo, (1620) Geographos, (3103) Eger, (161989) Cacus and (25143) Itokawa. The Vera C. Rubin Observatory, delivering sparse photometry for millions of asteroids, should turn a dozen into thousands.

Top Shapes, Moons, and an Asteroid Caught Coming Apart

The predicted end state is visible in the field. (101955) Bennu (490 m mean diameter, 4.296 h) and (162173) Ryugu (896 m, 7.63 h) are diamond-shaped rubble piles with sharp equatorial ridges, imaged by OSIRIS-REx from December 2018 and Hayabusa2 from June 2018. Both returned samples — 5.4 g from Ryugu on 6 December 2020, 121.6 g from Bennu on 24 September 2023 — confirming they are unconsolidated aggregates of primitive carbonaceous material. Ryugu's shape does not match its present slow spin; Watanabe and colleagues (2019) argued the ridge was built when it rotated near 3.5 hours. The same morphology appears on (65803) Didymos, whose 780 m primary spins at 2.26 h — essentially at the barrier.

Most such bodies have companions, the second prediction. Pravec and collaborators (2006) found that 15 ± 4 percent of near-Earth asteroids larger than 300 m are binaries, with rapid primaries clustered against the barrier and secondaries typically 20–50 percent of the primary diameter. Kevin Walsh, Derek Richardson and Patrick Michel (2008) simulated it directly: spin an aggregate past critical and it sheds a stream of particles that re-accretes into a satellite. Didymos and its 151 m moon Dimorphos — struck by NASA's DART on 26 September 2022, shortening the mutual orbital period by about 33 minutes, and to be surveyed by ESA's Hera from December 2026 — are the archetype. Asteroid pairs on nearly identical heliocentric orbits show the same fingerprint: exactly the mass-ratio-versus-spin relation rotational fission predicts (Pravec et al. 2010).

The disruption itself has been photographed. P/2013 R3 was resolved by Hubble between October 2013 and February 2014 into at least ten fragments drifting apart at only 0.2–0.5 m s−¹, each with its own dust tail; David Jewitt's team ruled out an impact (far too slow) and sublimation (far too weak), leaving rotational disruption. The 4 km asteroid (6478) Gault grew two dust tails in 2018–2019 and was then found to rotate in 2.49 hours — close enough to the barrier that intermittent shedding is the natural reading.

YORP and the Things It Is Confused With

Yarkovsky is not YORP. Both come from photon recoil, but Yarkovsky is a net force that shifts the orbit while YORP is a net torque that shifts the spin. Yarkovsky also needs a thermal lag: the afternoon side runs hotter than the morning side, so the recoil is not radial and pushes the orbit outward or inward depending on spin direction. Remove thermal inertia and Yarkovsky vanishes while the YORP spin torque survives; and Yarkovsky scales as 1/D against YORP's 1/D², so YORP wins decisively at small sizes. It was measured on (6489) Golevka by radar in 2003 and on Bennu, drifting inward at 284 metres per year.

The two are coupled, and the coupling made YORP famous. Because YORP drives obliquity toward 0° or 180°, where Yarkovsky drift is maximal, it sharpens the delivery of main-belt fragments into the ν₆ and 3:1 resonances that feed the near-Earth population — and it explains the roughly 2:1 excess of retrograde rotators among NEAs, which drift inward toward ν₆. The same pairing produced the Slivan states: Stephen Slivan (2002) found Koronis-family members with clustered spin periods and obliquities, later explained as YORP-driven capture into a spin-orbit resonance.

BYORP (Ćuk and Burns 2005) is the same recoil acting on a tidally locked satellite, changing the mutual orbit on 10⁴–10⁵-year timescales. Direct radiation pressure acts through a point near the centre of figure and barely torques anything; with Poynting-Robertson drag it instead dominates the lives of dust grains. Outgassing torques change comet spins far faster — Rosetta watched 67P shift its rotation period by of order twenty minutes across the 2015 perihelion — but need volatiles. Tidal disruption at a planet's Roche limit, as with Comet Shoemaker-Levy 9 at Jupiter in 1992, is a different failure mode entirely: gravity gradients, not sunlight.

Where the Theory Gets Uncomfortable

The awkward result is Thomas Statler's (2009): the YORP torque is not a well-behaved function of an asteroid's global shape. Because the effect is a small residual left after enormous cancellation, unresolved topography matters out of all proportion: a single boulder a few percent of the body's radius can change the predicted torque by tens of percent, or reverse its sign — the difference between an asteroid that disrupts in a million years and one that spins down into tumbling.

Bennu proved the point unhelpfully: the spin-up measured from its lightcurves was not what the pre-encounter radar shape model predicted, and the far better OSIRIS-REx model narrowed the gap without cleanly closing it. Cotto-Figueroa, Statler, Richardson and Tanga (2015) drew the consequence — real spin evolution is a random walk rather than a monotonic ramp, because each avalanche or shifted boulder resets the torque. YORP is self-limiting, and bodies linger near the barrier far longer than a naive ramp-to-fission picture suggests.

Open questions remain:

• Why is every detection a spin-up? Symmetry says about half of all bodies should be decelerating, yet every confirmed detection is an acceleration. Selection effects explain part of it; tangential YORP, nearly always positive, may explain the rest.
• How much cohesion is really there? Sánchez and Scheeres estimate 10–100 Pa from van der Waals forces and 1950 DA appears to need ~64 Pa — yet the OSIRIS-REx sampling event on 20 October 2020 sank far deeper than expected, implying Bennu's surface is effectively cohesionless.
• Shed, fission, or crumble? Whether a supercritical rubble pile leaks material from a widening equator, splits into two comparable masses, or fails internally without visible mass loss depends on interior structure we cannot yet observe. Hera at Didymos offers the first interior sounding of a small binary.
• How often? Turning a handful of caught-in-the-act disruptions into a rate needs survey completeness for faint, briefly active objects — something Rubin is about to change dramatically.

YORP compared with the other feeble forces that reshape small bodies

Process	What it changes	Physical driver	Typical magnitude / timescale
YORP	Spin rate and spin-axis obliquity	Recoil of re-emitted thermal infrared from an asymmetric shape	Doubles the spin of a 0.1–1 km NEA in ~10⁵–10⁷ yr
Yarkovsky effect	Orbital semimajor axis	Same photon recoil, but a thermal lag biases it along the orbit	Bennu drifts −284 m per year; scales as 1/D
BYORP	Mutual orbit of a binary	Radiation recoil on a tidally locked satellite	Expands or shrinks the mutual orbit in ~10⁴–10⁵ yr
Direct radiation pressure	Orbits of very small bodies	Momentum of absorbed and reflected sunlight	~4.5 × 10−⁶ Pa at 1 au; a detectable orbital perturbation only for metre-scale bodies, and stronger than solar gravity only for micron-sized grains
Outgassing torque	Spin of active bodies	Sublimating ice acting as a rocket	Comet 67P shifted its period by ~20 min in one perihelion passage
Collisions	Spin, shape, survival	Impacts by other asteroids	Sets the spin distribution above ~40 km diameter