Pebble Accretion

The timescale problem that pebble accretion solves

For decades the standard story of planet formation had a hole in the middle of it. Dust grains stick together into pebbles; pebbles somehow become kilometre-scale planetesimals; planetesimals collide and merge into planetary embryos; embryos sweep up the remaining planetesimals to build cores; and a core heavier than about ten Earth masses captures a runaway envelope of gas to become a giant planet. Every link in that chain has been studied to death. The problem was the timing of the fourth step. Beyond about five astronomical units, the place where Jupiter and Saturn actually formed, classical planetesimal accretion — embryos growing by the gravitational capture of other kilometre-scale bodies — is simply too slow. It needs roughly ten million years or more to assemble a ten-Earth-mass core. But protoplanetary gas disks disperse in three to ten million years. The gas was gone before the core was finished. Giant planets, on paper, could not form in time.

Pebble accretion is the resolution. Instead of waiting for an embryo to laboriously sweep up scattered planetesimals from its own narrow feeding zone, it lets the embryo feed on the centimetre-to-metre pebbles that drift inward through the entire disk — and, crucially, lets aerodynamic gas drag do the catching. The result is a core that grows from roughly a Ceres mass to ten Earth masses in under one million years at a few AU. The gas is still there. The giant planet has time to form. The hole in the middle of the story is filled.

How it works: drag turns the core into a net

The entire mechanism rests on one fact: a pebble feels the gas, and a planetesimal does not. A kilometre-wide planetesimal has so much inertia per unit drag area that the gas barely touches its orbit on any timescale that matters; it accretes only by gravity, capturing the tiny fraction of bodies that happen to pass within a few times its physical radius. A centimetre pebble is the opposite. Gas drag acts on it within roughly an orbital period, so when a pebble's trajectory carries it past a growing core, the drag bleeds off the relative velocity that would otherwise have flung it back out on a hyperbolic flyby. Robbed of that energy, the pebble cannot escape. It spirals down and is accreted.

The consequence is dramatic. A core accreting pebbles does not have a capture radius set by its physical size or even by its gravitational reach for fast encounters — it has a capture radius set by how far away a pebble can be and still have its excess speed dissipated before it leaves. In the strongly settled regime that radius is essentially the Hill radius, the distance at which the core's gravity dominates over the star's tidal field. For a growing core at a few AU the Hill radius is hundreds to thousands of times the core's physical radius, so the effective capture cross-section is enlarged by factors of 10⁴ to 10⁶ in area. The core becomes an aerodynamic net spanning its whole Hill sphere, catching nearly every pebble the disk sends through it.

The disk obligingly sends a lot of pebbles. Because the gas in a protoplanetary disk is partly supported by its own pressure gradient, it orbits very slightly slower than the Keplerian speed — a sub-Keplerian headwind of typically tens of metres per second. Pebbles, which want to orbit at the full Keplerian speed, plough into this headwind, lose angular momentum, and spiral inward. This radial drift carries pebbles from the outer disk past every embryo on their way toward the star. The core does not have to go find its food; the disk delivers a continuous inward flux of pebbles past its door, and the net catches them.

Worked example: building Jupiter's core at 5 AU

Put numbers on it. Consider an embryo of mass M = 10⁻³ M⊕ (about a Ceres mass) orbiting at a = 5 AU in a disk around a Sun-like star. The Hill radius is

R_Hill = a · (M / 3 M_star)^(1/3)
       = 5 AU · (10⁻³ × 3×10⁻⁶ / 3)^(1/3)
       ≈ 5 AU · 1.0×10⁻³
       ≈ 5×10⁻³ AU ≈ 7.5×10⁵ km

Compare that with the embryo's physical radius. A Ceres-mass rocky body has a radius of about 500 km. So the Hill radius is already ~1500 times the physical radius. In the Hill (shear) regime the core sweeps pebbles from across a band of order R_Hill wide, so the capture cross-section is larger than the geometric cross-section by roughly (R_Hill / R_phys)² ≈ 2×10⁶. That single factor is the whole game.

Now the accretion rate. In the 2D Hill regime the rate at which a core sweeps up a midplane layer of pebbles of surface density Σ_peb is approximately

dM/dt ≈ 2 R_Hill · Σ_peb · v_shear
      where v_shear ≈ Ω · R_Hill  (Keplerian shear across the Hill scale)

For a plausible pebble surface density of Σ_peb ≈ 1 g/cm² at 5 AU, an orbital frequency Ω = 2π/(11.9 yr), and the Hill radius above, this delivers a growth rate that, integrated as the Hill radius itself grows with M^(1/3), drives the core upward by orders of magnitude. Detailed integrations (Lambrechts & Johansen 2012; Bitsch et al. 2015) give the headline result: starting from a streaming-instability seed, a core reaches roughly 10 Earth masses in a few hundred thousand to one million years at 5 AU — and continues to about 20 Earth masses, the pebble isolation mass, before it shuts itself off. Run the same calculation with kilometre planetesimals instead of pebbles and the cross-section collapses by six orders of magnitude; the growth time balloons past 10 Myr, longer than the gas disk survives. The contrast is the entire reason the theory exists.

Regimes: Bondi, Hill, 2D and 3D

Pebble accretion is not one rate but a sequence of regimes the embryo climbs through as it grows. They are distinguished by how the core's gravity compares with the pebble's approach speed, and by whether the pebble layer is thinner or thicker than the core's capture scale.

• Bondi (drift) regime. For small embryos the relevant scale is the Bondi radius R_B = GM/Δv², where Δv is the headwind-driven drift speed of the pebbles. The core captures pebbles that pass within R_B. Accretion rises steeply with core mass here — this is the slow, rate-limiting branch the embryo must escape.
• Hill (shear) regime. Once the core is massive enough that R_B exceeds R_Hill, the Bondi description breaks down and the core captures pebbles from across its entire Hill sphere, fed by Keplerian shear. Growth is fast. Most of a giant-planet core's mass is laid down here.
• 2D vs. 3D. If the settled pebble layer is thinner than the core's capture radius, accretion is two-dimensional and reads off the pebble surface density (fastest). If turbulence puffs the pebble layer thicker than the capture radius, accretion becomes three-dimensional and reads off the volume density — slower, because the core sees only the slice of pebbles within its capture height.

The practical upshot is that turbulence is the enemy of efficiency: a quiescent, settled midplane pebble layer makes the net work at full 2D strength, while a stirred-up disk dilutes the flux into 3D and slows the build.

When the net switches off: pebble isolation mass

Pebble accretion is self-limiting, and the limit is elegant. As the core grows it perturbs the surrounding gas. Once it passes a threshold mass it generates a partial gap, and just outside its orbit the gas pressure develops a local maximum — a pressure bump. Pebbles drift toward pressure maxima, not toward the star, so at that bump the inward-drifting pebbles pile up and stall. The flux that was feeding the core is cut off. The core has reached its pebble isolation mass.

For a disk at 5 AU the pebble isolation mass is roughly 20 Earth masses, scaling with the cube of the disk's aspect ratio (Lambrechts et al. 2014; Bitsch et al. 2018). This is wonderfully convenient, because the critical core mass needed to trigger runaway accretion of a gaseous envelope is about 10 Earth masses. So the very event that halts solid growth — the pressure bump that starves the core of pebbles — also reduces the rate at which accretion energy is dumped into the core's nascent atmosphere, letting that atmosphere cool, contract, and run away into a gas giant. The same physics that builds the core then hands off cleanly to envelope capture.

Pebble accretion versus planetesimal accretion

Property	Pebble accretion	Planetesimal accretion
Building block size	cm–dm pebbles (Stokes ~ 0.01–1)	1–100 km planetesimals
Coupling to gas	Strong — drag bends trajectories	Negligible — gravity only
Effective cross-section	∝ Hill radius (100–1000× R_phys)	∝ gravitational focusing of R_phys
Material source	Whole disk, via inward radial drift	Local feeding zone, ~10 Hill radii
Core build time at 5 AU	< 1 Myr to 10 M⊕	> 10 Myr (often too slow)
Self-limiting?	Yes — pebble isolation mass	Isolation mass set by feeding-zone depletion
Key references	Ormel & Klahr 2010; Lambrechts & Johansen 2012	Safronov 1969; Wetherill & Stewart 1989

The two are not mutually exclusive — modern population-synthesis models use both. Streaming-instability planetesimals seed the embryos, planetesimal collisions contribute in the dense inner disk, and pebble accretion dominates the rapid build of cores in the outer disk where it must.

Observational status

The strongest support comes from ALMA. Its millimetre-wave continuum images trace exactly the pebble-sized grains the theory invokes, and they reveal that real disks — TW Hydrae, HL Tau, the dozens of disks in the DSHARP survey — are full of drifting pebbles organised into bright rings and dark gaps. The rings are most naturally explained as pebbles trapped at pressure bumps just outside the orbits of forming planets: the observable signature of the pebble isolation mechanism. Closer to home, the isotopic split between carbonaceous and non-carbonaceous meteorites (Kruijer et al. 2017) is best read as evidence that Jupiter's core grew fast enough, within the first million years, to act as a barrier dividing the disk's pebble reservoir into two populations — direct circumstantial support for rapid pebble accretion in our own Solar System. A clean, direct image of pebbles accreting onto a specific embryo remains the prize the next generation of high-resolution disk imaging is chasing.

Common pitfalls and misconceptions

• "A pebble is defined by its size." It is defined by its aerodynamics — a Stokes number near unity. The physical size that corresponds to that varies from millimetres to metres depending on disk density and grain porosity. Calling it a centimetre pebble is shorthand, not a definition.
• "Gas drag slows accretion." The opposite, for pebbles: drag is precisely what bleeds off the relative velocity and lets the core capture material it would otherwise miss. Drag is the enabling mechanism, not an obstacle.
• "The core gravitationally pulls the pebbles in." Gravity alone would only deflect a fast-moving pebble onto a hyperbolic flyby. It is the combination of gravity to deflect and drag to dissipate that achieves capture. Remove either and the net fails.
• "Pebble accretion runs away without limit." It is self-terminating at the pebble isolation mass, when the core perturbs the gas into a pressure bump that blocks the incoming pebble flux. That ceiling is a feature, not a bug — it sets the core mass that giant planets start from.
• "It works the same everywhere." Efficiency depends on location, turbulence, pebble Stokes number, and whether an upstream core has already starved the inner disk. It is most powerful in the outer disk, which is exactly where giant-planet cores must form fast.
• "It replaced planetesimal accretion." It complements it. Streaming-instability planetesimals provide the seeds; pebble accretion grows them. Both rely on the same gas-drag physics at different scales.

Quantitative analysis: why the cross-section explodes

To see precisely why pebble accretion outpaces planetesimal accretion, compare the two capture cross-sections. For a planetesimal of physical radius R_phys, gravitational focusing gives an effective capture radius

R_capture² ≈ R_phys² · (1 + v_esc²/v_rel²)

where v_esc is the core's surface escape speed and v_rel the relative encounter velocity. Even with strong focusing, R_capture is at most a small multiple of R_phys for the random velocities typical of a planetesimal swarm, and crucially the focusing weakens as the swarm gets dynamically hot. For pebbles, by contrast, the relevant comparison is set not by the physical radius but by the timescales of drag and encounter. A pebble passing the core at impact parameter b is captured if gas drag dissipates its excess speed within the time the core's gravity acts on it. Working this out, the effective accretion radius in the Hill regime is

R_acc ≈ R_Hill · (St / 0.1)^(1/3)        (St ≲ 0.1, settled)
R_acc ≈ R_Hill                            (St ~ 0.1–1, optimal)

So for Stokes numbers near the sweet spot the capture radius is the full Hill radius. Now the ratio that decides everything:

(R_acc / R_capture)²  ≈  (R_Hill / R_phys)²  ≈  (1500)²  ≈  2×10⁶

at our 5 AU example. The pebble net is two million times larger in area than the planetesimal-focusing cross-section. Feed that into dM/dt = (cross-section) × (flux) and the growth time drops from >10 Myr to <1 Myr. That single factor — gas drag enlarging the capture cross-section from the physical radius to the Hill radius — is what lets pebble accretion build a giant-planet core before the disk gas disappears, and why it has become the default mechanism in modern planet-formation theory.