Circumbinary Planet

What "circumbinary" really means

A binary star system is two stars bound together gravitationally, orbiting a common center of mass — their barycenter. Surrounded by them, but far enough out that the pair looks like a single object, a circumbinary planet traces one big orbit around the whole binary. The technical name for this geometry is a P-type orbit ("P" for planetary, indicating the planet sees the binary as one combined mass). The complementary case, in which a planet orbits one star of a wide binary while the second star sits much farther out as a distant companion, is an S-type orbit (S for satellite). The Alpha Centauri system, where any planet around α Cen A or α Cen B would be in an S-type configuration, is the textbook example. Tatooine, with two suns close together and the planet outside both of them, is a P-type — a circumbinary planet.

The distinction matters because the physics is completely different. An S-type planet feels one nearby star's gravity strongly and the distant companion as a slow tidal perturber. A P-type planet feels the inner pair as a single mass on long timescales but as a rapidly varying two-body potential on short timescales — and that variability is what makes circumbinary orbits dangerous.

The Holman-Wiegert stability limit

The defining theoretical result for circumbinary planets is the 1999 numerical study by Matthew Holman and Paul Wiegert. They integrated test-particle orbits around binaries of varying mass ratio μ = M₂ / (M₁ + M₂) and eccentricity e_bin, and asked: for each combination, what is the smallest planetary semi-major axis a_crit at which orbits stay bound for the full integration time (10⁴ binary periods)? Their fitting formula is

a_crit / a_bin = 1.60 + 5.10 e_bin − 2.22 e_bin²
              + 4.12 μ − 4.27 e_bin μ − 5.09 μ²
              + 4.61 e_bin² μ² + 4.61 μ² e_bin

For typical Kepler circumbinaries — eclipsing binaries with e_bin in the 0.0–0.2 range and μ near 0.3 — the predicted a_crit lands between 2.5 and 3.0 a_bin. Inside that radius, the planet's orbit becomes chaotic on Myr timescales and the planet is ejected; outside it, orbits are stable for the age of the galaxy.

Almost every known circumbinary planet sits remarkably close to its own a_crit — typically at 1.0 to 1.5 times the critical radius. This "piling up" near the stability limit is not an accident: it is one of the strongest pieces of evidence that these planets formed far out in their natal disks and migrated inward, halting just when further migration would have destabilised them.

Kepler-16b — the first Tatooine

On 15 September 2011, the Kepler team announced Kepler-16(AB) b, the first confirmed planet around a main-sequence eclipsing binary. The light curve of the binary, KIC 12644769, showed both stellar eclipses (deep and periodic) and a second, third type of dip that recurred neither on the binary period nor on simple harmonics. Modelling the dips required a third body in a P-type orbit. Follow-up radial-velocity work tightened the parameters.

The binary consists of a K-dwarf (Kepler-16A, mass ≈ 0.69 M☉) and an M-dwarf (Kepler-16B, mass ≈ 0.20 M☉) in a 41-day, nearly circular orbit. The planet, Kepler-16b, is roughly Saturn-mass (M ≈ 0.33 M_Jup, R ≈ 0.75 R_Jup), and orbits the pair every 229 days at a ≈ 0.7 AU — well outside the predicted a_crit ≈ 0.36 AU. The Kepler team and the press immediately christened it Tatooine after the desert world with two suns in Star Wars, a comparison George Lucas himself endorsed at the press conference.

Kepler-16b transits both stars during its orbit, producing characteristic deep + shallow eclipse pairs that depend on which star is foreground at the moment of crossing. Each transit lands at a slightly different time and depth because the binary continues to whirl while the planet's transit window is open — exactly the timing irregularity that ruled out an ordinary single-star planet hypothesis.

The known catalog

The Kepler mission produced the bulk of the confirmed circumbinary catalog. TESS has added a few more since 2020, and a handful of pre-Kepler candidates from eclipse timing variation studies have been folded in. The current confirmed roster looks roughly like this:

System	Discovery	P_bin (d)	P_planet (d)	R_planet	Notes
Kepler-16	2011	41.1	229	0.75 R_J	First Tatooine; Saturn-mass
Kepler-34	2012	27.8	289	0.76 R_J	Twin G stars; eccentric binary
Kepler-35	2012	20.7	131	0.73 R_J	Saturn-mass; near-Earth size HZ
Kepler-38	2012	18.8	106	0.39 R_J	Neptune-class; close to a_crit
Kepler-47	2012 / 2019	7.45	49 / 187 / 303	0.27 / 0.41 / 0.66 R_J	Three planets; one in HZ
Kepler-64 (PH1)	2012	20.0	138	0.55 R_J	Quadruple system; planet hunters discovery
Kepler-413	2014	10.1	66	0.39 R_J	Precessing orbit; transits intermittent
Kepler-453	2015	27.3	240	0.55 R_J	Marginally inside HZ
Kepler-1647	2016	11.3	1107	1.06 R_J	Longest period; Jupiter-size
TOI-1338 / BEBOP-1	2020 / 2023	14.6	95 / 215	0.59 R_J / Neptune	First TESS circumbinary; two planets

The masses span Neptune-class to slightly super-Jupiter; nothing convincingly Earth-sized has been confirmed yet. The orbital periods cluster between 50 and 300 days, with Kepler-1647 b a striking outlier at three years. All of them are P-type. The total reaches about 13 confirmed planets, though the number depends on how you count systems with multiple planets and disputed timing-variation detections.

Detection — transits with a twist

Three signatures combine to confirm a circumbinary planet from photometry alone.

• Planetary transits across either star. Each transit is a measurable dip, often shallower than the stellar eclipses but distinctly shaped. When the planet crosses the smaller, fainter secondary star, the dip is deeper relative to that star's contribution; when it crosses the larger primary, the absolute drop is smaller but easier to detect.
• Transit timing variations (TTVs). The mid-time of each planetary transit drifts from a fixed cadence by hours or days, because the geometry of "planet between us and star" depends on where the stars are in their own orbit at the moment of crossing. A naive single-star fit produces transit residuals of order P_bin / P_planet × some geometric factor — a smoking-gun pattern that no single-star planet can mimic.
• Eclipse timing variations (ETVs). Symmetrically, the planet's gravity perturbs the binary's own eclipse times. ETVs on a planet-induced cadence (typically the planet's orbital period) provide a model-independent mass estimate; the photometric transit depths give the radius. Combining the two yields a density.

The Kepler-16 discovery used all three. Importantly, ETVs alone — without transits — can flag circumbinary candidates, but they are notoriously prone to mimicry by stellar activity, apsidal motion, and unmodelled hierarchical companions. A bona fide circumbinary detection currently requires a transit.

Kepler-47 — three planets, one binary

If Kepler-16b proved that circumbinary planets exist at all, Kepler-47 proved that planetary systems can form there. Announced in 2012 (planets b and c) and updated in 2019 (planet d), Kepler-47 hosts three confirmed planets around a Sun-like primary and a much fainter M-dwarf in a 7.45-day orbit.

• Kepler-47 b — innermost; 49-day orbit; R ≈ 3.0 R_⊕ (mini-Neptune scale).
• Kepler-47 d — middle; 187-day orbit; R ≈ 7.0 R_⊕ (Neptune-class). Sits inside the system's habitable zone.
• Kepler-47 c — outermost; 303-day orbit; R ≈ 4.7 R_⊕.

(The labels b, c, d follow discovery order, not orbital order — Kepler-47d, discovered later, ended up between b and c in real space.) The system is dynamically tightly packed and demonstrates that the inner truncation of the disk by the binary does not prevent multiple planets from forming further out. It also gives a rare statistical handle on disk dispersal timescales: forming three planets in a circumbinary disk requires the disk to survive long enough at the cool outer radii for accretion to complete.

The breathing habitable zone

The classical habitable zone (HZ) for a single star is the annulus where the time-averaged stellar irradiance allows liquid water on a rocky surface — typically 0.95 to 1.7 AU around a Sun-like star. For a binary, you cannot simply add the two stars' luminosities and treat them as a point source, because the planet's distance to each star changes with binary phase. Two regimes matter:

• Mean irradiance. For a circular binary, the time-averaged flux at planetary distance r is exactly the same as from a single star of luminosity L₁ + L₂ at the same r (the binary motion averages out at large planetary distance). So the HZ has nearly the same boundaries as for the combined-luminosity single-star equivalent.
• Modulation. The instantaneous flux varies as the stars approach and recede from the planet's location. For a 41-day binary like Kepler-16, the modulation amplitude at the planet is at the 10–20% level — comparable to Earth's annual variation due to orbital eccentricity. The HZ "breathes": its boundaries oscillate inward and outward on the binary period.

Detailed climate-model studies (Haghighipour & Kaltenegger 2013; Kane & Hinkel 2013; Forgan 2014) find that circumbinary HZs are real and finite, slightly broader than equivalent single-star HZs, but with extra climate variability that may rule out narrow, knife-edge habitability conditions in favour of more buffered atmospheres.

Formation — how to build a planet on a fault line

The fundamental challenge is that the same binary perturbations that destabilise mature planets also wreck the planet-forming disk. Three things go wrong inside the stability radius:

1. Tidal truncation. The binary clears the inner disk out to roughly 2 to 3 a_bin — the same scale as a_crit. The warm, dense, terrestrial-planet-forming region of a normal protoplanetary disk simply does not exist around a tight binary.
2. Eccentric forcing. Even outside the truncation radius, the binary pumps eccentricities into nearby planetesimals to ~0.05–0.1. Mutual collision velocities rise above the escape speed of the planetesimal itself, so collisions shatter bodies instead of merging them. Pebble accretion is less affected than planetesimal accretion, which probably explains why circumbinaries are gas-giant-dominated.
3. Strong spiral waves. Resonant gravitational coupling launches spiral density waves into the disk, sweeping up dust into pressure bumps and modulating accretion rates onto growing protoplanets.

The favoured solution since Pierens & Nelson (2007) is form farther out, migrate inward. Planets accrete in the calmer outer disk (r ≳ 5–10 a_bin), where conditions resemble a normal protoplanetary disk. Type-I or Type-II gas-disk migration then drags them inward. The migration halts when the planet reaches the inner edge of the gas disk — where the gas density drops sharply at the tidally truncated boundary — because the asymmetric torque from the truncated disk no longer drives net inward motion. That edge sits just outside a_crit, which is precisely where observed circumbinary planets pile up. The story is internally consistent and has become the standard picture.

Worked example: where would Kepler-16b sit?

Take the Holman-Wiegert formula and plug in Kepler-16:

M_A = 0.69 M☉,  M_B = 0.20 M☉  →  μ = M_B / (M_A + M_B) = 0.225
a_bin = 0.224 AU
e_bin = 0.159

a_crit / a_bin = 1.60 + 5.10(0.159) − 2.22(0.159)²
              + 4.12(0.225) − 4.27(0.159)(0.225) − 5.09(0.225)²
              + (higher-order terms, small)
            ≈ 1.60 + 0.811 − 0.056 + 0.927 − 0.153 − 0.258
            ≈ 2.87

a_crit ≈ 2.87 × 0.224 AU ≈ 0.64 AU

The observed planet sits at a = 0.7048 AU, just 1.10 × a_crit — within 10% of the stability boundary. This is the canonical "pile-up": the planet halted exactly where the inner truncation of the disk forced its migration to stop.

Applying the same exercise to Kepler-34 (μ ≈ 0.5, e_bin ≈ 0.52, a_bin ≈ 0.23 AU) gives a_crit ≈ 0.84 AU; the planet sits at 1.09 AU = 1.30 × a_crit. Kepler-1647 (a_bin ≈ 0.13 AU) has a_crit ≈ 0.40 AU but the planet sits at 2.72 AU = 6.8 × a_crit — an outlier, far from the pile-up, which is itself interesting and may indicate either truncated migration or in-situ formation of a Jupiter analog at large radius.

Future detections — TESS, PLATO, and beyond

The Kepler era ended in 2018; the modern circumbinary search runs on TESS and ground-based BEBOP. TESS's all-sky cadence is shorter than Kepler's, so it favours short-period planets, but its sky area is far larger — the trade-off has so far produced TOI-1338b (= BEBOP-1b) in 2020 and a second planet in the same system (BEBOP-1c) confirmed via radial velocities in 2023. The ESA PLATO mission, slated to launch in 2026, will revisit the long-cadence Kepler-style strategy on a fresh patch of sky and is expected to substantially expand the circumbinary catalog, including potentially Earth-sized planets in the habitable zone of solar-type binaries — the obvious next milestone.

The longer-term goal is statistical: with enough confirmed systems, we can ask whether planet occurrence rates depend on binary period, mass ratio, or eccentricity, and whether the pile-up at a_crit is universal. Current upper limits on Earth-sized circumbinary planets are still loose; ruling them out, or detecting them, would be major science.

Common pitfalls

• Confusing P-type and S-type. Tatooine is P-type (planet around both stars). A planet "in a binary system" is ambiguous and may be S-type (around one star only). The dynamical physics is completely different.
• Treating ETV alone as a discovery. Eclipse timing variations can be mimicked by stellar activity, apsidal motion, or an unseen hierarchical triple companion. A genuine circumbinary planet detection currently requires a photometric transit signature; ETV is corroboration, not proof.
• Forgetting that a_crit depends on e_bin and μ. The often-quoted "2.5 a_bin" is the round number for circular equal-mass binaries; the full Holman-Wiegert formula is required for any real system, and a_crit can run 1.6 to 5 × a_bin depending on parameters.
• Assuming circumbinary planets are rare because we found few. The detection geometry is severe: the planet's plane must align with the binary's, the binary itself must be eclipsing for most current surveys to target it, and transit cadences must catch a relatively long-period planet during the mission. The intrinsic occurrence rate may be ≳ 10% of binaries, comparable to single-star planet rates.
• Misapplying habitable-zone formulae. A single-star HZ at L_total = L₁ + L₂ is only the leading-order approximation. For tight or eccentric binaries, the modulation amplitude becomes a first-class effect, and climate stability must be assessed via dedicated 1D or 3D atmospheric models, not the textbook Kasting boundaries.